Avascular necrosis (AVN) of the femoral head is an increasingly common cause of musculoskeletal disability, and it poses a major diagnostic and therapeutic challenge. Although patients are initially asymptomatic, avascular necrosis (AVN) of the femoral head usually progresses to joint destruction, requiring total hip replacement (THR), usually before the fifth decade (see the images below). In fact, 50% of patients with avascular necrosis experience severe joint destruction as a result of deterioration and undergo a major surgical procedure for treatment within 3 years of diagnosis. Femoral head collapse usually occurs within 2 years after development of hip pain. [1, 2, 3]
The incidence of avascular necrosis (AVN) is increasing. The causes include greater use of exogenous steroids and an increase in trauma. [4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19] In 54-80% of renal transplant recipients in whom AVN is detected with plain radiographs, the disease is bilateral.
It is estimated that almost 10% of the nearly 500,000 THRs performed each year in the United States are intended to treat AVN; at a cost of more than $1 billion, THRs performed to treat avascular necrosis (AVN) of the femoral head constitute approximately 25% of the total national costs for THR. Trauma is the most common cause of avascular necrosis; however, nontraumatic avascular necrosis (AVN) is commonly bilateral and occurs in younger persons. In addition, nontraumatic bilateral AVN usually occurs at different times and progresses at different rates in different hips.
Treatment of avascular necrosis (AVN) has been facilitated by the adoption of an international classification system, by effective early diagnosis using magnetic resonance imaging (MRI), and by more aggressive surgical management; nevertheless, no universally satisfactory therapy has been developed, even for early disease.
Because measures to preserve the joint are associated with better prognoses when the diagnosis of avascular necrosis (AVN) is made early in the course of the disease and because the results of joint replacement therapy are poorer in younger age groups than in older patients, it is critical to diagnose this condition as early as possible to prevent or delay progression of the disease.
Avascular necrosis (AVN) is characterized by areas of dead trabecular bone and marrow extending to involve the subchondral plate. The anterolateral aspect of the femoral head, the principal weightbearing region, is typically involved, but any region of the femoral head may be involved. In the adult, the involved segment usually never fully revascularizes, and collapse of the femoral head usually occurs sometime after avascular necrosis (AVN) is detected radiographically.
The femoral head is the most vulnerable site for the development of avascular necrosis (AVN). The site of necrosis is usually immediately below the weightbearing articular surface of the bone (ie, the anterolateral aspect of the femoral head). This is the site of greatest mechanical stress.
Elderly persons are at decreased risk for developing avascular necrosis (AVN). Fat cells become smaller in elderly persons. The space between fat cells fills with a loose reticulum and mucoid fluid, which are resistant to avascular necrosis (AVN). This condition is termed gelatinous marrow. Even in the presence of increased intramedullary pressure, interstitial fluid is able to escape into the blood vessels, leaving the spaces free to absorb additional fluid.
To understand the changes of avascular necrosis (AVN) on radiologic studies such as computed tomography (CT) scanning and MRI, it is necessary to understand the anatomy of the hip and to understand the vascular anatomy of the hip in children.
The hip is a ball-and-socket joint. The acetabulum, which provides bony coverage of 40% of the femoral head, has a horseshoe-shaped lunate surface. The femoral head is round and smooth in all imaging planes. The fovea capitis, a small depression on the medial femoral head, is the site of attachment of the ligamentum teres (see the image below).
The principal sources of blood flow to the femoral head are the lateral epiphyseal vessels (LEVs). LEVs are branches of the posterior superior retinacular vessels (PSVs), which are themselves branches of the medial femoral circumflex artery; the medial femoral circumflex artery is a branch of the profunda femoris artery (see the following 2 images). The PSVs run along the posterior-superior aspect of the femoral neck under the synovial membrane. They are extraosseous in location and give rise to the LEV (see the following 2 images). 
The LEV enters the femoral head within a 1-cm-wide zone between the cartilage of the femoral head and the cortical bone of the femoral neck. They supply the lateral and central thirds of the femoral head (see the following 2 images). When patent, the artery of the ligamentum teres (ALT) supplies the medial third of the femoral head.
Branches of the LEVs and the ALT anastomose in the junction of the central and medial third of the femoral head. The thickest part of the articular cartilage of the femoral head is located along the posterior-superior aspect and measures 3 mm in diameter. It thins to 0.5 mm along the peripheral and inferior margins.
Blood supply in children
In children 4-7 years of age, the vascular anatomy of the proximal femur is in a transitional stage of development. The ALT does not penetrate the epiphysis of the femoral head until age 9 or 10 years. The medial circumflex artery, a branch of the profunda femoris artery, penetrates into the femoral proximal metaphysis but is prevented from passing into the femoral epiphysis by the growth plate. The blood supply to the femoral head is especially vulnerable during this time.
Anatomy on CT scans
Physiologic thickening of bone trabeculae in the center of the femoral head is present and appears similar to a star, which is termed the asterisk sign (see the image below). The configuration is related to the stress of weightbearing.
Sclerotic raylike branches of the star usually extend to the upper surface of the femoral head (see the image below). A dense line, extending from the lateral to the medial portion of the mid femoral head, represents the fused epiphysis.
Anatomy on MRI
Fatty marrow is present in the femoral capital epiphysis and the greater trochanter of all individuals older than 2 years. Fatty marrow has high signal intensity on T1-weighted images (T1WIs) and T2-weighted images (T2WIs) (see the images below). Hematopoietic marrow, when present, is found in the femoral neck, the intertrochanteric region, and the acetabulum. It has low signal intensity on T1WIs and high signal intensity on T2WIs (see the images below).
The medullary cavity contains prominent vertically orientated linear striations of low signal on all imaging sequences extending from the inferolateral aspect to the superomedial aspect of the femoral head. These represent the weightbearing trabeculae and are analogous to the asterisk sign seen on CT scans (see the 2 images above). The medullary cavity is surrounded by a sharply marginated line of low signal intensity, which represents the cortex of the bone. Cortex and trabeculae have weak MRI signal intensity because of a low concentration and decreased mobility of hydrogen ions (see the 2 images above).
A thin line of high signal intensity, which represents the articular cartilage, surrounds the outer margin of the femoral head. A curvilinear low-signal line, representing the physis, crosses the marrow of the femoral neck laterally to medially (see the 2 images above). The medullary cavity of the iliac bone, adjacent to the acetabulum, is of slightly lower and less homogeneous signal intensity than the femoral head (see the 2 images above).
Sequelae of avascular necrosis
Avascular necrosis (AVN) progresses from minimal to more severe disease to mechanical failure.
Minimal avascular necrosis
If the vascular area is small and is not adjacent to an articular surface, the patient may be asymptomatic; healing may occur spontaneously, or the disease may remain undetected or be discovered incidentally during workup for other conditions.
More severe avascular necrosis
Once AVN develops, repair begins at the interface between viable bone and necrotic bone. Ineffective resorption of dead bone within the necrotic focus is the rule. Dead bone is reabsorbed only partially. Reactive and reparative bone is laid down on dead trabeculae, resulting in a sclerotic margin of thickened trabeculae within an advancing front of hyperemia, inflammation, bone resorption, and fibrosis. The incomplete resorption of dead bone has a mixed sclerotic and cystic appearance on radiographs. Necrosis and repair are ongoing in various stages of evolution within a single lesion.
Mechanical failure of trabecular bone at the interface between dead and viable bone may exacerbate avascular necrosis (AVN). In the subchondral region, such microfractures do not heal because they occur within an area of dead bone. Progression of the microfractures results in a diffuse subchondral fracture, seen radiographically as the crescent sign (see the first image below). Following subchondral fracture and progressive weightbearing, collapse of the articular cartilage occurs (see the second through fifth images below). Continued fracture, necrosis, and further weightbearing may progress to degenerative joint disease (DJD) and joint dissolution (see the second image and last 2 images below).
This section will discuss various imaging modalities. For more detailed information, please see their respective sections in this article.
MRI is the most sensitive means of diagnosing avascular necrosis (AVN). This imaging modality provides the criterion standard of noninvasive diagnostic evaluation and is more sensitive than CT scanning or planar scintigraphy. In addition, MRI is much more sensitive than plain film radiography for detecting avascular necrosis (AVN). However, low-field magnets (0.1 Tesla [T]) are not as sensitive for diagnosing avascular necrosis (AVN).
7-T hip MRI showed comparable results in hip joint imaging compared with 3 T, with slight advantages in contrast detail (cartilage defects) and fluid detection at 7 T when accepting image degradation medially. Image homogeneity of 7 T compared with 3 T (3.9-4.0 for all sequences) was degraded, especially in TSE sequences at 7 T through signal variations (7 T: 2.1-2.9). 
MRI is indispensable for the accurate staging of avascular necrosis (AVN), because images clearly depict the size of the lesion, and gross estimates of the stage of disease can be made (see Radiograph for the radiographic and radiologic staging systems for AVN). MRI allows sequential evaluation of asymptomatic lesions that are undetectable on plain radiographs.  MRI facilitates better response to treatment because, with the use of MRI, avascular necrosis (AVN) is diagnosed at an earlier stage, and therapeutic measures are more successful the earlier they are begun.
MRI does not employ ionizing radiation—a factor that is especially important in the growing skeleton—and it is accepted widely and is easy to perform. MRI is also capable of imaging in multiple planes (ie, axial, sagittal, coronal, or any variation thereof), demonstrates superior soft-tissue resolution, and has high spatial and contrast resolution, allowing evaluation of morphologic features.
MRI may help guide interventional procedures such as core decompression, may demonstrate response of the femoral head to treatment, and may detect the joint effusions and bone edema that often accompany avascular necrosis (AVN). MRI is a noninvasive means of evaluating articular cartilage congruity, and it allows sequential evaluation of asymptomatic lesions that are undetectable on plain radiographs. [22, 23, 24, 25, 26]
Early MRI detection and closed bone graft epiphysiodesis (CBGE) may mitigate the effects of AVN after slipped capital femoral epiphysis (SCFE). Seventeen patients (17 hips) had a scheduled MRI between 1 and 6 months from initial surgery. Six hips diagnosed by MRI received surgical intervention (4 CBGE, 1 free vascularized fibula graft, and 1 repinning due to screw cutout) at a mean of 4.1 months (range, 1.3 to 7.2 mo) postoperatively. None of the 4 patients treated with CBGE within 2 months postoperatively progressed to stage IVC AVN. 
Similar sensitivity and specificity were found between low-field MRI and planar radionuclide bone scanning. At high magnetic field strength, MRI has a higher sensitivity than radionuclide scanning. Using a 1.5-T magnet, Beltran et al reported 88% sensitivity, 100% specificity, and 94% accuracy with MRI and 78% sensitivity, 75% specificity, and 76% accuracy with bone scintigraphy.  The high sensitivity of MRI has been confirmed by others.
In another study, MRI performed at 0.6 T and single-photon emission CT (SPECT) bone imaging using technetium-99m (99m Tc) methylene diphosphonate (MDP) were similarly effective in diagnosing avascular necrosis (AVN).  MRI had a sensitivity of 87% and a specificity of 83%; whereas SPECT scanning had a sensitivity of 91% and a specificity of 78%. Both were more effective than planar bone scintigraphy, which had a sensitivity of 83% and a specificity of 83%.
MRI was also more effective than SPECT for diagnosing cases of avascular necrosis (AVN) of the hip in which pain was absent. MRI detected avascular necrosis (AVN) in 10 of 15 patients, but SPECT scanning detected AVN in only 5 of 15 patients.
Using receiver operating characteristic (ROC) curves, MRI was better than CT scanning by more than 2 standard errors and better than radionuclide scanning by more than 3 standard errors in helping diagnose early avascular necrosis (AVN).
Magnetic resonance perfusion imaging has been able to identify significant differences between avascular necrosis, bone marrow edema, and subchondral insufficiency fractures of the proximal femur, particularly regarding maximum enhancement values (Emax), slope (Eslope) and time to peak (TTP). Diffusion weighted imaging of bone marrow of the proximal femur did not show significant differences in the same study. 
Gadolinium-enhanced perfusion MRI (pMRI) after closed reduction/spica casting for developmental dysplasia of the hip (DDH) has been suggested as a potential means to identify and avoid avascular necrosis (AVN) by helping the surgeon evaluate femoral head vascularity. 
Single-photon emission CT scanning
SPECT scanning provides images of the radioactivity within the target organ in 3 dimensions. With this modality, overlying and underlying areas of radioactivity may be separated into sequential tomographic planes, thus providing increased image contrast and improved lesion detection and localization, as compared with planar scintigraphy. SPECT scanning eliminates radioactivity resulting from hyperemia about the hip joint and from the underlying acetabulum and adjacent bladder. SPECT scanning is used as an alternative to MRI when MRI cannot be performed or when the results of MRI are indeterminate. [31, 32, 33]
Initially, SPECT images reflect vascular integrity; early in the disease, SPECT scans may demonstrate an avascular focus; such findings are missed with MRI unless contrast is used. Collier et al found a sensitivity of 85% for SPECT scanning.  With triple-head high-resolution SPECT scanning, Lee et al reported a sensitivity of 97%. 
Planar radionuclide imaging, bone scintigraphy using pinhole collimation, and planar scintigraphic imaging using quantitative bone scanning are briefly discussed below.
Collier reported a sensitivity of 55% with planar radionuclide imaging for avascular necrosis (AVN).  However, bone scintigraphy equipped with a pinhole collimator has greater sensitivity for diagnosing avascular necrosis (AVN) than bone scintigraphy using a high-resolution parallel-hole collimator.
The pinhole collimator is a conical collimator with a small circular aperture (3-5 mm) that produces an inverted image of the object in a manner analogous to photographic cameras. The image obtained is magnified, allowing better visualization of small structures and improving detection of scintigraphic abnormalities. The pinhole collimator optimizes resolution in the evaluation of circumscribed areas, and the acquisition time is only 15 minutes, compared to up to 45 minutes for SPECT scanning. The technique is an alternative to MRI when MRI cannot be performed or when MRI results are not clear-cut.
Planar scintigraphic imaging using quantitative bone scanning provides physiologic data that cannot be obtained with other modalities, including MRI; for example, this technique allows quantification of uptake in the perfusion and static phases. However, correct computer programming is required. [36, 37, 38, 39]
In one study, F-18 fluoride PET/CT showed good agreement with MRI in the initial diagnosis of AVN and was better than MRI in detecting early disease. MRI was 96.5% sensitive, 100% specific, and 98.03% accurate, while PET/CT was 100% sensitive, 100% specific, and 100% accurate in diagnosing AVN. 
In a comparative study of technetium-99m-methylene diphosphonate (99mTc-MDP) SPECT/CT versus planar bone scintigraphy (BS) for diagnosis of AVN, SPECT/CT was found to be superior to planar BS and SPECT alone. The diagnostic accuracy of planar BS, SPECT, and SPECT/CT was 67%, 78%, and 95%, respectively. Planar BS was found to have the lowest sensitivity (75%) and specificity (40%), whereas SPECT/CT had the highest sensitivity (98%) and specificity (87%). 
The high spatial resolution and contrast resolution of CT scanning allow analysis of morphologic features (see the images below), and . The sensitivity of CT scanning in detecting early avascular necrosis (AVN) is 55%, which is similar to the sensitivity of planar nuclear medicine imaging. Thus, CT scanning is more appropriate in evaluating the extent of involvement, such as subchondral lucencies and sclerosis during the reparative stage, before the onset of femoral head collapse and superimposed degenerative disease.
CT scanning is better able to help define the extent of disease at stages II and higher than MRI and plain film radiography. CT scanning enables detection of subchondral or cancellous fractures and collapse, especially when multiplanar reconstruction is used. This information is essential for planning treatment (see the third and fourth images above).
Plain film radiography
Although unable to detect disease of stage 0 or 1, plain film radiography may be helpful in assessing flattening of the femoral head and associated degenerative changes (see the images below). See Radiograph.
Limitations of techniques
The limitations of the imaging techniques discussed above are briefly reviewed.
Bone biopsy analysis has been rarely reported to be positive when MRI findings are normal. MRI cannot be performed in patients who have cardiac pacemakers or when intracranial clips are present, nor can MRI be performed in patients who have claustrophobia. In addition, problems related to malpositioning may lead to misrepresentation. In children, slight pelvic obliquity may cause the normal dark-appearing growth plate to appear in the same axial cut as the contralateral bright-appearing epiphysis; in such cases, the normal growth plate may appear to be abnormal on MRI.
Children may require sedation as a result of the long imaging times that MRI requires. It may be difficult to detect avascular necrosis (AVN) after surgery to repair a hip fracture because of the presence of orthopedic hardware, which creates significant image distortion. Marrow cells are more resistant to ischemia than hematopoietic cells or osteocytes. Because MRI images reflect changes within marrow fat signal intensity, MRI findings of avascular necrosis (AVN) may not be seen for up to 5 days after the ischemic event, until the marrow fat cells have died. In this situation, contrast-enhanced MRI is needed.
SPECT scanning demonstrates poor spatial resolution. Artifacts from the bladder are frequently encountered; these artifacts may obscure the photon-deficient region of the femoral head. A number of techniques, such as the use of multihead cameras with shorter acquisition times that improve resolution and increase sensitivity, have been advocated, but none has gained universal acceptance.
SPECT imaging requires a cooperative patient who must remain immobile for up to 45 minutes of acquisition time, thus this modality is difficult to use in children because of the necessity to remain motionless for long periods of time. Children may require sedation. In addition, diagnosing Legg-Calve-Perthes (LCP) in small children may be difficult because of the small size of the femoral epiphysis and associated bladder artifacts.
Planar scintigraphy demonstrates poor spatial resolution (see the following images). The ring of increased activity reflecting hyperemia in the early stages and bone healing later obscures the photon-deficient necrotic center within the femoral head, which is indicative of avascular necrosis (AVN). The site may show a uniform high level of activity, making it impossible to distinguish avascular necrosis (AVN) from other causes of increased activity, such as osteoarthritis, fracture, and inflammatory arthritis. A cold spot in the femoral head is highly specific but not sensitive for diagnosing avascular necrosis (AVN).
Artifacts from radioactivity in the bladder are frequently encountered, obscuring the photon-deficient region. Entities causing increased uptake about the hip joint, such as arthritis and inflammatory disease, may obscure the photopenic necrotic focus within the femoral head. Results can be judged only by comparison with the other hip and may be of little use in the presence of bilateral involvement.
Although CT scanning may delineate subtle alterations of bone density when plain radiograph findings are normal, MRI and SPECT scintigraphy are much more sensitive for evaluating early manifestations of the disease, such as bone marrow edema. CT scans are insensitive for detecting stage 0 and 1 avascular necrosis (AVN), but they are excellent for detecting femoral head collapse, early degenerative joint disease (DJD), and the presence of loose bodies.
CT scanning may improve the accuracy of radiographic staging using thin-slice thicknesses of 1 mm or less and by incorporating multiplanar reconstruction. In one study, 30% of hips with stage 2 (precollapse) avascular necrosis (AVN), evaluated with plain film radiography, had stage 3 disease when evaluated with CT scans.
Plain film radiography
Using plain film radiography, the sensitivity for detecting early stages of the disease is as low as 41%. Plain film does not detect stage 0 and 1 avascular necrosis (AVN). A delay of 1-5 years may occur between the onset of symptoms and the appearance of radiographic abnormalities. Normal radiographic findings do not necessarily mean that disease is not present.
Demineralization of the femur may be detected, and the disease may be suggested only after bone resorption has occurred. If early diagnosis is needed for the prompt initiation of therapy, more sensitive imaging methods (ie, MRI) must be used, especially in patients who are at increased risk for AVN.
Differential diagnosis and other problems to be considered
Bone metastases should be considered in the differential diagnosis as well as transient osteoporosis. Other problems to be considered when viewing findings on plain film radiographs, bone scintigraphs, CT scans, and MRIs are as follows:
Plain film radiograph
Transient osteoporosis of the hip
Advanced DJD (see the following image)Frogleg lateral view of the right hip in a patient with avascular necrosis shows the crescent sign, indicating subchondral fracture. Therapeutic interventions are less likely to halt progression of the disease once this sign appears. The frogleg lateral view is better than anteroposterior (AP) projection for demonstrating this sign, because the anterior and posterior margins of the acetabulum on the AP projection are superimposed over the superior portion of the femoral head, the usual location of the sign.
Plasma cell myeloma
Bone marrow edema syndrome
Plasma cell myeloma
Transient osteoporosis of the hip
Transient bone marrow syndrome
Epiphyseal stress fracture
Failure to diagnose such a potentially devastating condition as avascular necrosis (AVN) in a young age group has the potential for serious medical-legal repercussions. Malpractice settlements reflect compensation for a lifetime of a potentially compromised lifestyle with much morbidity. Such settlements also reflect the cost of potential joint replacement and prosthesis failure.
Diagnosing AVN as early as possible is imperative for a greater chance of success of conservative treatment. Patients who are at high risk must be screened using MRI. Normal radiograph findings do not mean a normal hip. Failure to pursue this condition with more aggressive imaging in a high-risk population can potentially lead to medical malpractice.
MRI has replaced bone marrow pressure, venography, and bone biopsy. These are invasive procedures that were highly sensitive and specific for diagnosing avascular necrosis (AVN). They should be performed only when a high index of suspicion is present and all tests are equivocal.
Ficat and Arlet radiographic staging system for AVN
A staging system using radiographic findings was developed by Ficat and Arlet and has been used widely for treating avascular necrosis (AVN).  However, their system has been supplanted by the classification system of Steinberg et al (see below), which incorporates MRI and scintigraphic findings. 
Stage 0 (preclinical and preradiologic)
Avascular necrosis (AVN) can be suggested only if it has already been diagnosed in the contralateral hip.
Stage 1 (preradiologic)
Since the advent of MRI, stage 1 avascular necrosis (AVN) is defined by normal findings on radiographs and positive findings on MRI or bone scintigraphy. Stage 1 represents the early resorptive stage. Late in this stage, plain radiographs may show minimal osteoporosis and/or blurring and poor definition of the bony trabeculae. Osteoporosis appears when one third of the mineral content of bone has been lost.
Stage 2 (reparative)
Stage 2 avascular necrosis (AVN) represents the reparative stage before flattening of the femoral head occurs and may extend for several months or longer. Demineralization is now evident; it is the first manifestation of the reparative stage, represents resorption of dead bone, and may be generalized or patchy or appear in the form of small cysts within the femoral head.
Patchy sclerosis represents apposition of new bone on dead trabeculae and appears after demineralization develops, usually in the superolateral aspect of the femoral head (see the images below). However, patchy sclerosis usually coexists with demineralization, appearing as alternating regions of increased density and increased lucency. On radiographs, patchy sclerosis appears as increased density and may be diffuse, focal, or in a linear arc, which is concave superiorly. The pattern demonstrates alternating areas of lucency and sclerosis.
Stage 3 (early collapse of the femoral head)
In stage 3 avascular necrosis (AVN), a linear subcortical lucency, representing a fracture line, is present immediately beneath the articular cortex and may extend into the articular cartilage at the superolateral aspect of the femoral head. This is termed the crescent sign and is best demonstrated on a frogleg view (see the images below). The subarticular cortex may remain attached to the cartilage and is separated from the underlying femur by soft tissue, termed the eggshell sign. The femoral head initially preserves its round appearance, but later, it demonstrates collapse. This may be indicated by joint-space widening.
Stage 4 (progressive degenerative disease)
Further flattening of the femoral head occurs with loss of its smooth convex contour in stage 4 avascular necrosis (AVN) (see the following image). Ultimately, the superior femoral fragment, representing the articular surface and the immediate subchondral bone, may become separated from the underlying femoral head or depressed and compacted into the femoral head. Fragments of bone and cartilage may separate from the underlying femur, roam freely within the hip joint, and become loose bodies.
Severe collapse and destruction of the femoral head leads to progressive degenerative joint disease (DJD) with joint-space narrowing, marginal osteophyte formation, and subchondral cyst formation. Subchondral cysts can usually be differentiated from the alternating sclerosis and the lucency of the reparative stage of avascular necrosis (AVN).
Atypical radiographic findings are seen in 18% of patients and those on steroid therapy. These findings consist of early joint-space narrowing, often before the appearance of the crescent sign. Unless the physician holds a high index of suspicion for avascular necrosis (AVN), an incorrect diagnosis of osteoarthritis will be made. Furthermore, signs of bone repair (sclerosis) may be absent; the first radiologic manifestations may be the subchondral lucency representing fracture of the dead bone.
Atypical findings occur because bone formation is decreased in the presence of normal bone resorption; in this situation, increased density within the femoral heads usually is a result of flattening from fracture and compression of the femoral head.
Steinberg et al's staging system for AVN
Steinberg et al proposed a 6-stage classification system based on that of Ficat and Arlet and included radiologic clinical classification findings  :
This stage is both preclinical and preradiologic. Most patients with stage 0 disease are identified when imaging is performed to evaluate avascular necrosis (AVN) in the contralateral hip or to exclude other diseases. Abnormal MRI findings, normal radiographic findings, and normal bone scan findings are features of stage 0.
Stage 1 avascular necrosis (AVN) demonstrates normal radiographic findings or shows minimal demineralization or blurred trabeculae. Pain in the anterior groin or thigh is common. Limited range of motion (ROM) in the hip may be present. Abnormal bone scan findings, mild groin pain, and normal radiographic findings are features of stage 1.
This stage shows diffuse or localized areas of sclerosis, lucencies, or both within the femoral head. Clinical signs persist or worsen. Osteoporosis, groin pain, and mottled sclerotic and/or cystic areas are features of stage 2 avascular necrosis (AVN).
Stage 3 avascular necrosis (AVN) is characterized by the crescent sign (subchondral fracture). A crescent line, pain with subchondral fracture activity, and no femoral head flattening are features of stage 3.
This stage demonstrates marked collapse and fracture involving the articular surface. Segmental flattening of the femoral head demonstrates an out-of-round appearance. Thus, Segmental flattening, pain with femoral head activity, no acetabular involvement, and normal joint space are features of stage 4 avascular necrosis (AVN).
Stage 5 avascular necrosis (AVN) is characterized by the development of DJD. Thus, joint space narrowing, resting pain, and acetabular degeneration (DJD) are features of stage 5
Advanced staging of AVN using plain radiography
The Association Research Circulation Osseous (ARCO) of the Toulouse, France-based Association Internationale de Recherche sur la Circulation Osseuse has proposed a further classification of the various stages of avascular necrosis (AVN), which incorporates the percentage (area) of involvement of the femoral head and the location of the lesion. In addition, the extent of the AVN lesion is an important determinant of both clinical and radiologic outcomes.
Three types of involvement have been identified, mild, moderate, and severe. In mild disease less than 15% of the femoral head involvement is noted, which is less likely to demonstrate radiographic progression or require hip prosthesis. In moderate disease, the femoral head involvement ranges from 15% to 30%. In severe, disease, the femoral head involvement is greater than 30%. Moderate and severe involvement are more likely to progress radiographically to degenerative disease and to require hip prosthesis placement.
Degree of confidence
The degree of confidence of certain radiologic features of avascular necrosis (AVN) are briefly discussed below.
Demineralization is a nonspecific finding seen in a large number of different diseases. Such a finding needs further evaluation using MRI to evaluate for avascular necrosis (AVN).
Alternating areas of lucency and sclerosis
This feature is characteristic for stage 2 disease avascular necrosis (AVN). Rarely, this finding is confused with entities such as chondroblastoma, a radiolucent cartilaginous tumor that contains calcium and is located in the epiphyseal region. If there are questions concerning the presence of the disease, MRI is recommended. If not, treatment can be initiated.
DJD with degenerative spurring and joint space narrowing with subchondral cyst formation may mimic avascular necrosis (AVN). Subchondral cysts are usually immediately adjacent to areas of joint-space narrowing and osteophyte formation. MRI is usually diagnostic in problematic cases. Rarely, biopsy may be needed for differentiation.
Some radiologic features of avascular necrosis (AVN) may lead to false-positive findings.
Poorly defined radiolucent lesions may simulate the bone destruction seen in malignancy, osteomyelitis, and transient osteoporosis of the hip (TOP). For malignancy and osteomyelitis, the history may be helpful.
Demineralization can be seen in a number of different diseases, including TOP. TOP is self-limiting and resolves within 4-10 months. Radiologic resolution lags behind clinical improvement by 4-8 weeks, at which time radiographic findings revert to normal. The healing and reparative phase may mimic bone sarcoma.
The later stages of avascular necrosis (AVN), which are characterized by joint-space narrowing, articular cartilage destruction, and alternating areas of lucency and sclerosis within the femoral head, may mimic DJD with subchondral cyst formation (see the following images).
Sclerosis adjacent to an insufficiency fracture is an important differential, especially in patients who are osteopenic and are taking steroids.
CT scans do not demonstrate the early vascular and marrow abnormalities that result in osteonecrosis.  In fact, osteoporosis is the first visible CT scan sign of avascular necrosis (AVN). Later, the central bony asterisk is distorted, appearing as clumping and fusion of the peripheral asterisk rays. Clumping appears as spots or as hyperdense "roads" of various width (see the following image). This represents changes in the sclerotic interface between necrotic and viable bone and is analogous to the line of low signal surrounding the necrotic bone seen on MRI images.
Early signs are caused by microfractures resulting from reduced mechanical load of dead bone trabeculae, altering the shape of the asterisk and are related to new bone formation on the dead trabeculae. The lucent cystic region, representing the reparative zone, may be appreciated (see the image above).
Degree of confidence
Osteoporosis, whether diagnosed using plain film radiography or CT scanning, must be evaluated further, because it is present in a great number of diseases. MRI findings are usually diagnostic.
Unless the asterisk sign is appreciated, articular surface abnormalities may be interpreted as DJD. The lucency within the reparative zone may be confused with malignancy, infection, insufficiency fracture, or plasma cell myeloma.
Magnetic Resonance Imaging
When using MRI to evaluate avascular necrosis (AVN), the coronal plane is the most important imaging plane, and sagittal images may help eliminate partial-volume averaging, which is especially present on axial images. When the lesion is located anterosuperiorly, off-coronal images, angled toward the axial plane, may demonstrate better avascular necrosis (AVN). [45, 46, 47]
Because both hips are often involved in avascular necrosis (AVN), and the condition is silent early in the course of the disease, use of a body coil and a large field of view (30-40 cm) is necessary to image both hips simultaneously. Surface coils, including shoulder coils, flexible coils, and phased-array coils, may provide additional resolution for individual hip joints in selected patients. T1WIs and T2WIs are obtained in the coronal plane, 4-mm thick, with a 1-mm gap. Fast spin-echo (FSE) images with fat saturation may also be obtained.
Short tau inversion recovery (STIR) images provide excellent fat suppression and demonstrate areas of bone marrow edema (see the images below). STIR images may be obtained using FSE techniques with an echo train length of 8-16. This helps reduce lengthy imaging times associated with STIR imaging.
A frequency selective pulse may be added to suppress the fat signal. When applied, the inner bright line on T2WIs is visualized, but the dark outer peripheral band is not seen. Nevertheless, FSE T2WIs with fat suppression are useful in demonstrating the extent of marrow edema associated with AVN.
Rapidly acquired MRI sequences can reliably reveal the presence of avascular necrosis (AVN). These rapid screening sequences reduce or eliminate artifacts caused by patient motion. Coronal 2-dimensional (2-D), fast, low-angle shot (FLASH) T1WIs are performed using repetition time (TR,) 174.9 milliseconds (ms); echo time (TE), 4.1 ms; flip angle, 70°; 4-mm slice thickness with 20% interslice gap; matrix, 172 × 256; number of signal acquisitions, 1; and imaging time, 39 seconds (s). Axial fat-suppressed FSE T2WIs are performed using TR, 3500 ms; TE, 138 ms; echo train length, 29; 6-mm slice thickness with 25% interslice gap; matrix, 116 × 256; number of signal acquisitions, 1; and imaging time, 16 s.
Chemical shift imaging may be used to detect premature fatty marrow conversion associated with avascular necrosis (AVN). Fatty and hematopoietic marrow and the distribution of water within the ischemic focus can be differentiated on fat-selective and water-selective images.
Gradient-echo images are not as sensitive for fluid within reparative tissue but they can demonstrate joint effusions, subchondral fluid, and changes in the contour of the articular cartilage.
NOTE: Screening for avascular necrosis using T1WI only reduces specificity, may fail to identify a transchondral fracture, and may not help diagnose other diseases in which clinical presentation may mimic AVN, including transient osteoporosis.
Khanna et al described a limited magnetic resonance examination using coronal T1WIs, in which MRI introduction earlier in the diagnosis of femoral head osteonecrosis, as well as its more widespread use in patient care, may be allowed due to the time and and potential cost reduction achieved with a limited examination.  The investigators also reported excellent agreement between the full and screening MR examinations for both the detection of and determining the extent of osteonecrosis.
Only 1 case of avascular necrosis (AVN) was missed in 29 patients, and the time required for the limited exam was 10 minutes relative to the 30 minutes required for the full examination.  Without calculating the professional component, relative costs of the screening evaluation was $104, whereas and the full assessment was $312. However, other diseases causing hip pain (eg, myositis, greater trochanteric bursitis, labral cysts, and fractures) that were located distant from the femoral heads were missed when T2WIs were not obtained.  Perhaps there may be a role for the limited exam in following up bone marrow edema in asymptomatic patients.
A peripheral band of low signal is present in the superior portion of the femoral head outlining a central area of bone marrow. This is considered to represent the reactive interface between the necrotic and reparative zones and extends to the subchondral bone plate (see the images below, 2 of which are followed by T2WI images in the same respective patients).
The inner border of the peripheral band demonstrates high signal. This may represent chemical shift artifact because the position of the signal changes when the phase and frequency directions are changed. This is termed the double-line sign and is pathognomonic for avascular necrosis (AVN) (see the first image below). It is present in 80% of cases. This is not demonstrated well on FSE T2WIs because of the increased signal intensity of fat, present on this sequence, which obscures the bright inner line (see the second image below). To compensate, a frequency selective pulse is added to suppress the signal from fat. If fat suppression is used, the dark outer peripheral band of avascular necrosis (AVN) is not demonstrated well, in contrast to the inner high-signal band visualized on this sequence. The outer low-signal ring represents the interface of repair tissue with the necrotic zone.
Use of contrast enhancement
If intravenous contrast is used to supplement the MRI examination, areas of decreased enhancement indicate early avascular necrosis (AVN) despite normal findings on pre-enhancement images.  Contrast enhancement is useful for distinguishing viable from nonviable trabeculae and marrow. Nonviable tissue does not enhance after contrast administration. Enhancement of the low-signal band on T1WIs corresponds to the reparative zone.
Atypical MRI findings
Avascular necrosis (AVN) occasionally appears as an area of abnormal signal involving the femoral head, neck, and intertrochanteric region. It is characterized by decreased signal on T1WIs and increased signal on T2WIs without the focal lesions that are pathognomonic for avascular necrosis (AVN). This is termed the bone marrow edema pattern, as the signal characteristics are consistent with increased free water or edema within the normal fatty marrow of the proximal femur. This may reflect early edema before the onset of focal abnormalities and may indicate the time period between cell death and development of a significantly large reactive interface, which is recognizable as avascular necrosis (AVN) on MRIs.
Differentiation between transient and irreversible AVN lesions
When using subchondral marrow changes on T2WIs or contrast-enhanced T1WIs to differentiate transient from irreversible avascular necrosis (AVN) lesions, the absence of low-signal subchondral lesions and subchondral deformities in the presence of the bone marrow edema pattern represents transient osteoporosis. Areas of low signal intensity in the subchondral region and contour deformities of the femoral head are typical of avascular necrosis (AVN).
Associated MRI findings
Fatty conversion of marrow is a prerequisite for developing avascular necrosis (AVN) of the femoral head, a finding that may help identify populations at increased risk for developing the disease. Subchondral fractures may appear as a low signal intensity gap on T1WIs; on T2WIs, they can appear as regions of high signal intensity, representing fluid within the fracture line. Joint effusions are present in 50% of patients (see the following images).
AVN classification per central avascular segment signal alterations
The avascular necrosis (AVN) lesion is classified into 4 types according to alterations in the central avascular segment signals on MRI.
Central osteonecrotic focus signal analogous to that of fat are noted. Increased signal is demonstrated on T1WIs, and intermediate to high signal is demonstrated on T2WIs (see the image below).
The presence of central osteonecrotic focus signal analogous to that of blood is observed. Increased signal is demonstrated on both T1WIs and T2WIs (see the following images).
Central osteonecrotic focus signal analogous to that of fluid is present. Decreased signal is demonstrated on T1WIs, and increased signal is demonstrated on T2WIs (see the 2 images above, and the images below).
The presence of central osteonecrotic focus signal analogous to that of fibrous tissue is noted. Decreased signal is demonstrated on both T1WIs and T2WIs.
MRI staging, symptoms, and prognostic correlations
Correlations between MRI staging and radiographic staging, MRI class and clinical symptoms, and MRI findings and prognosis are outlined in this section. (See Steinberg et al's Staging System for AVN in Radiograph.)
Correlation between MRI and radiographic staging
MRI classes A and D showed the best correlation with radiographic staging.
Approximately 50% of radiographic stage 1 and 83% of radiographic stage 2 lesions demonstrated MRI class A signal pattern.
In those femoral heads complicated by fracture (radiographic stages 3 and 4), only 14% were MRI class A and 43% were MRI class B.
MRI class B and C lesions correlate poorly with radiographic staging.
Correlation between MRI class and clinical symptoms
Of patients with MRI class A lesions, 54% were asymptomatic.
Of patients with MRI classes B and C lesions, 11% were asymptomatic.
Of patients with MRI class D lesions, 67% were asymptomatic.
Correlation between MRI findings and prognosis
MRI classes, unlike radiographic stages, have little predictive value regarding prognosis for femoral head collapse. Entirely circumscribed avascular necrosis (AVN) that did not extend to the subchondral margin had a good outcome, independent of the overall size of the lesion. The percentage of the weightbearing surface occupied by the lesion was the most reliable factor in predicting outcome.
Basing outcome on overall extent of involvement of the femoral head on MRI is controversial. Lafforgue et al evaluated 3 different means of determining femoral head involvement and found that the percentage of weightbearing femoral cortex involved with avascular necrosis (AVN) was the most reliable parameter in determining outcomes.  Beltran et al found that femoral head collapse occurs in most patients with a large area of avascular necrosis (AVN) before the appearance of a subchondral fracture, even if core decompression is performed.  Using MRI, the investigators determined femoral head collapse did not occur when less than 25% of the weightbearing surface was involved.
Femoral head collapse tendency and MRI lesion size
Tendency toward femoral head collapse in relation to avascular necrosis (AVN) lesion size as demonstrated on MRI is in agreement with the quantitative radiographic staging of Steinberg et al (see Steinberg et al's Staging System for AVN in Radiograph). [52, 53, 54, 55, 48, 56, 57, 50, 58, 59, 60, 61, 62, 63, 64, 65, 66] Small lesions confined to the medial anterosuperior portion of the femoral head tended not to collapse over a 28-month follow-up period. However, more extensive lesions collapse, with a 50% collapse rate within 12 months. Shimizu et al found a 74% rate of femoral head collapse by 32 months if the region of avascular necrosis (AVN) involved more than two thirds of the weightbearing surface area. 
Degree of confidence
MRI has a high sensitivity in the diagnosis of bone marrow abnormalities; the sensitivity of MRI in the diagnosis of avascular necrosis (AVN) is 85-100%. MRI has 97% sensitivity in distinguishing a hip with AVN involvement from a normal hip. In differentiating avascular necrosis (AVN) from non-AVN disease of the femoral head, MRI demonstrates a sensitivity of 98% and a specificity of 85%. Before femoral head collapse, the specificity is 75-100%. After femoral head collapse, the sensitivity is 100%.
In a retrospective analysis of the possible predictive ability of contrast-enhanced MRI for avascular necrosis (AVN) after closed reduction for developmental dysplasia of the hip (DDH) in infants, Tiderius et al suggested that gadolinium-enhanced MRI provides accurate anatomic assessment of a closed reduction in DDH as well as information about femoral head perfusion that may be predictive for future AVN.  Multivariate logistic regression indicated that a global decreased enhancement was associated with a significantly higher risk of developing avascular necrosis (AVN).  However, the authors noted that further investigation is required before perfusion information can be used for routine clinical use.
Bone marrow edema
This is a nonspecific finding seen in avascular necrosis (AVN) and other conditions that may progress to frank AVN (see the images below).
Transient osteoporosis of the hip (TOP)
This is a self-limiting condition characterized by osteoporosis of the femoral head and, occasionally, the femoral neck. TOP resolves over a 4- to 10-month period, and it does not progress to avascular necrosis (AVN). TOP often can appear on both sides of the hip joint, differentiating it from avascular necrosis (AVN). Similar findings can develop in the contralateral hip or other joints, in which case it is termed regional migratory osteoporosis. TOP occurs in patients without the risk factors for avascular necrosis (AVN).
Transient bone marrow syndrome
This condition is similar to TOP, but osteoporosis is never present radiographically. Symptoms are self-limiting, and it occurs in patients who have no risk factors for AVN.
This condition usually is self-limiting and resolves over time. If bone marrow edema is present on MRI, plain radiography is obtained. If plain film findings are normal, radiography should be repeated within 4-6 weeks. If osteoporosis is detected, it is believed to represent TOP. If osteoporosis is absent, patients may be placed into groups with and without high risk factors for developing avascular necrosis (AVN). Patients with factors indicating a low index of suspicion can be treated conservatively, but plain radiography and MRI follow-up imaging should be performed. High-risk patients should be considered candidates for surgical intervention.
An area of increased signal on T2WIs in the subchondral zone may represent fracture or edema. To accurately stage the disease, CT scans are helpful in differentiating the 2 conditions. Yeh et al determined that the accuracy of routine MRI was not satisfactory when compared with CT scanning in identifying subchondral fracture in avascular necrosis (AVN).  A false-positive diagnosis was not uncommon. Therefore, the investigators suggested the interpretation of routine MR imaging readout should be guarded. 
False-negative MRI diagnosis may be related to the use of T1WIs only. These images are less sensitive to detecting the bone marrow edema pattern of early avascular necrosis (AVN). This is detected better using T2WIs or STIR images. 
SPECT scanning and perfusion and static planar radionuclide imaging are discussed in this section.
A cold spot (photon-deficient region) within the femoral head is highly specific for avascular necrosis (AVN) and is the earliest scintigraphic evidence of this disease. The finding is usually seen 7-10 days after the ischemic event.
Over a period of weeks to months, increased uptake representing revascularization and repair surrounds and eventually replaces the region of photopenia. The central region of photopenia with surrounding zone of increased uptake is termed the doughnut sign.
Perfusion and static planar radionuclide imaging
Initially, uptake is decreased in the perfusion and static phases, which represents the early ischemic event. Later, uptake is decreased within the femoral head in the perfusion phase and increased around the cold region in the static phase. The latter represents the reactive zone around the infarcted segment. The increased uptake from the reparative zone eventually replaces the photopenic region.
Degree of confidence
A cold spot can be seen in other conditions, such as infection, metastasis, joint effusion, and plasma cell myeloma. Spencer et al reported that not all adults take up radiopharmaceutical agents in the femoral head. As a result, MRI is needed for confirmation. 
False-negative findings in planar scintigraphy
Hungerford reported false-negative bone scans in the hips of 14 of 27 patients, 13 of whom had bilateral disease.  In patients with bilateral involvement, the uptake, although symmetric, really is increased bilaterally. If the uptake is asymmetric, the side affected more severely makes the less-involved side appear falsely normal.
Later in the course of the disease, between the time of infarct and revascularization, the scan appears falsely normal in 6-10% of patients or it demonstrates a pattern of uptake that cannot be differentiated from DJD.
False-positive findings in planar scintigraphy
Decreased uptake, indicative of early avascular necrosis (AVN), can also be seen in infection, plasma cell myeloma, skeletal metastasis, hemangioma, and radiation therapy.
Increased uptake alone can be seen in arthritis, sympathetic dystrophy, malignancy, infection, transient osteoporosis of the hip (TOP), hemangioma, and insufficiency fractures. TOP can cause increased uptake on both sides of the hip joint.