Pneumonia is predominantly a clinical syndrome. [1, 2, 3] The classic etiologic agents of atypical pneumonia are Legionella species, Mycoplasma pneumoniae, and Chlamydia pneumoniae. Many other diseases, caused by various pathogens, should be considered in the differential diagnosis. Such etiologic agents include fungi, mycobacteria, parasites, and viruses (eg, influenza virus, adenovirus, respiratory syncytial virus, human parainfluenza virus, measles, varicella zoster, Hantavirus).
In immunosuppressed patients, outbreaks of isolated cases of respiratory virus infections with atypical presentations are reported. These infections can be severe and may have concomitant bacterial etiologies. In endemic areas, certain zoonotic infections should be considered when patients present with atypical pneumonia. Noninfectious etiologies must be considered in atypical and nonresolving pneumonias.
During the latter half of the 19th century, by which time physicians had embraced autopsy as an essential learning tool, pneumonia diagnoses were usually made post mortem. With the discovery of x-rays (1895), chest radiography became part of the routine evaluation of pneumonia in patients with suggestive signs and symptoms. Patients who presented with fever, shaking chills, and rust-colored sputum (which under examination showed gram-positive diplococci in chains) and whose chest radiographic findings were suggestive of pulmonary infection were considered to have typical pneumonia. Scans of atypical bacterial pneumonia are depicted below.
A phase of active hyperemia occurs, lasting approximately 24 hours before radiologic consolidation of the alveoli appears. This phase is characterized by engorgement of the arterial blood vessels. Edematous fluid, which may be seen in the alveolus, contains few exudative cells.
The next stage is referred to as red hepatization. Neutrophils and fibrin material fill the alveoli, and massive extravasation of red blood cells produces a homogeneous opacity.
The red hepatization phase is then followed by gray hepatization. Fibrin and exudative cells accumulate, appearing on radiographs as a clear zone adjoining the alveolar and acinar cells.
If the process extends to the pleural space, associated empyema may be present.
Legionella species are implicated in 2-15% of community-acquired pneumonia (CAP) cases.
These organisms usually cause a patchy, localized infiltrate in the lower lobes. Associated hilar adenopathy may be present. Pleural effusion is seen in up to 30% of cases. In rare instances, Legionella infection is associated with cavitation and a masslike appearance.
Radiologic resolution of Legionella pneumonia may take 6-12 months. Permanent residual fibrosis is observed in as many as 25% of patients. An early progression of infiltrates can occur despite clinical improvement. [5, 6, 7, 8, 9, 10, 11, 12, 13] Legionella pneumonia is depicted in the image below.
Chest radiographs cannot be used to distinguish nosocomial legionellosis (Legionnaires disease) from other pneumonias.
M pneumoniae is implicated in 2-30% of all cases of CAP. Mycoplasma pneumonia is usually mild and results in a rapid resolution of any radiologic findings. However, it tends to be more severe in patients with sickle cell anemia. Radiographic resolution in 40% of patients occurs in 4 weeks, and 80% of cases resolve by 8 weeks. Residual radiographic abnormalities are uncommon.
The infiltrates in Mycoplasma pneumonia can be unilateral, multilobar, or bilateral.  In about 20% of patients, pleural effusion or hilar adenopathy may be present. Mycoplasma pneumonia is depicted in the image below.
The infiltrates may be subsegmental or more extensive in elderly patients; pleural effusions are rarely seen. Chest radiographs show 50% resolution in 4 weeks. In 20% of cases, resolution takes longer than 9 weeks. Chlamydia pneumonia is depicted in the image below.
Degree of confidence
Radiologic findings alone are not reliable in differentiating pneumonia into typical or atypical forms. Therefore, the radiographic findings described above should be used along with clinical and laboratory data to narrow the possibilities.
Structural lung disease with abnormal lung parenchyma affects the pattern of infiltrates. In cases of severe emphysematous lung disease, clinicians may tend to underestimate the presence of infiltrates on chest radiographs.
Computed tomography (CT) scans are increasingly being used in clinical practice. Various authors have questioned CT scanning's usefulness in evaluating consolidations, suggesting that the value of CT in the diagnosis of pneumonia is limited to specific cases involving the following [15, 16] :
An indistinct, abnormal opacity on chest radiographs
Patchy, ground-glass, linear, or reticular opacities on chest radiographs
Possible pleural effusion
Neutropenia and fever of unknown origin (for which ultra–thin-section CT scanning may be helpful)
High-resolution CT findings in CAP
Tanaka et al compared high-resolution CT (HRCT) scan findings in CAP with pathologic findings and evaluated the role of HRCT scanning in differentiating between bacterial and atypical pneumonias in 32 patients with CAP (18 with bacterial pneumonia, 14 with atypical pneumonia). 
Bacterial pneumonia often resulted in airspace consolidation with a segmental distribution (72%) that typically occurred toward the middle and outer zones of the lungs. Atypical pneumonias included Mycoplasma and Chlamydia pneumonias, as well as influenza viral pneumonia. These conditions frequently caused a centrilobular shadow (64%), an acinar shadow (71%), and/or airspace consolidation (57%) and ground-glass attenuation (86%) with a lobular distribution. The lesions were often distributed to the inner, middle, and outer layers of the lung (86%). A CT scan depicting chlamydial pneumonia is shown in the image below.
Mild Legionella pneumonia may manifest with bilateral involvement of the lung parenchyma. Multiple segments are affected, and peripheral lung consolidation with ground-glass opacity and pleural effusion may be seen. With more severe infection, lung cavitation and bulging of the fissure have been reported. Residual lung parenchymal scarring can be found, even after the acute infection resolves.  A CT scan depicting Legionella pneumonia is shown in the image below.
Reittner et al examined 28 patients, identifying ground-glass attenuation in 24 (86%) and airspace consolidation in 22 (79%). In 13 of the latter 22 patients (59%), the areas of consolidation had a lobular distribution. Nodules were more common on HRCT scans (89%) than on radiographic images (50%), and in 24 of 28 patients (86%), the nodules had a predominantly centrilobular distribution on CT scans. Thickening of bronchovascular markings were more often found with CT scanning (82%) than with radiography (18%). 
Degree of confidence
Coinfection with several organisms is not uncommon. Underlying parenchymal lung abnormalities usually predispose patients to pneumonia. Therefore, in patients with pneumonia, the overall clinical and radiologic picture must be considered in place of an independent, dichotomous view. (See the image below.)
Magnetic Resonance Imaging
Robbins and Kumar described Legionella infection in an HIV patient who was found on MRI to have a lesion of the splenium of the corpus callosum. 
The literature suggests that ultrasonography can help in differentiating between consolidation and effusion. Consolidated lung tissue may appear as hypoechoic areas with blurred margins. The texture varies with the amount of aeration, being more heterogeneous with aeration and more homogeneous with dense consolidation. The literature also reports that ultrasonography may aid in the diagnosis of empyema and abscesses. However, the current authors believe that in clinical practice, ultrasonography's usefulness is limited to the identification and quantification of parapneumonic effusions. Once found, the area where an effusion occurs can be marked for subsequent diagnostic or therapeutic thoracentesis.