Compression of digital chest radiographs with a mixture of principal components neural network: evaluation of performance.

The performance of a new, neural network-based image compression method was evaluated on digital radiographs for use in an educational environment. The network uses a mixture of principal components (MPC) representation to effect optimally adaptive transform coding of an image and has significant computational advantages over other techniques. Nine representative digital chest radiographs were compressed 10:1, 20:1, 30:1, and 40:1 with the MPC method. The five versions of each image, including the original, were shown simultaneously, in random order, to each of seven radiologists, who rated each one on a five-point scale for image quality and visibility of pathologic conditions. One radiologist also ranked four versions of each of the nine images in terms of the severity of distortion: The four versions represented 30:1 and 40:1 compression with the MPC method and with the classic Karhunen-Loève transform (KLT). Only for the images compressed 40:1 with the MPC method were there any unacceptable ratings. Nevertheless, the images compressed 40:1 received a top score in 26%-33% of the evaluations. Images compressed with the MPC method were rated better than or as good as images compressed with the KLT technique 17 of 18 times. Four of nine times, images compressed 40:1 with the MPC method were rated as good as or better than images compressed 30:1 with the KLT technique.

Evaluation of the emphysematous patient.

Most of the patients with emphysema complain of dyspnea and become limited in their activities during the course of the disease. Dyspnea is probably due to the change in the configuration of the thorax which is secondary to hyperinflation. The investigation should include the radiologic quantification of the structural abnormalities of the lungs and the functional consequences of these changes. When volume reduction of the lung is considered, the effects of hyperinflation on the mechanic of breathing and the ventilatory response to exercise has to be investigated rigorously.

Minimizing the risk of Q fever in the hospital setting.

Q fever is caused by a rickettsial microorganism (Coxiella burnetii) harboured in sheep. The highest concentration of organisms are found in birth products. It is a very contagious organism which humans can contract by inhaling aerosolized organisms. Most commonly it leads to an acute 'flu-like illness. Rarely, chronic disease with endocarditis is fatal. Infected patients should be treated with tetracyclines or chloramphenicol. A number of outbreaks have been reported in hospital and research settings. Because of the fear of patients and staff contracting Q fever, Hospital Research Review Boards have increasingly resisted the presence of sheep in medical facilities. The authors have reviewed the circumstances leading to these outbreaks and believe researchers can minimize the risk of Q fever. The most important precautions are to use sheep only from Q fever controlled flocks and, depending on the nature of the research, only male sheep.

Automated in vivo quantification of emphysema.

A computer program to automatically determine lung volume and percentage of emphysema from computed tomographic (CT) chest sections was developed. To test it, the authors reviewed scans obtained in 89 patients. Any computer errors in identification of normal or emphysematous lungs were corrected manually by tracing the boundaries of the lungs or eliminating areas of nonemphysematous lung with a roller ball. The corrected values were compared with the uncorrected values. In 33 patients from the study group, lung volumes and the percentage of emphysema were calculated by using a currently available "voxel highlighting" program. Successful computerized analysis of entire lungs was performed in 1 minute, compared to more than 1 hour for manual analysis. Correlation was high (r = .99) between results obtained manually and those obtained with the computer. Although the difference between uncorrected and corrected values of the percentage of emphysema was statistically significant, this converted to a correction of only 0.1%. The computer program allowed quick, accurate, and reproducible quantification of emphysema.

The trachea: pathologic conditions and trauma.

A variety of common and uncommon conditions affect the trachea. Respiratory symptoms rarely occur until there is a 50% narrowing of the trachea. Chondromalacia, sometimes seen in conjunction with congenital tracheoesophageal fistula, can be identified with fluoroscopy. Patients with tracheal narrowing who undergo general anesthesia are at great risk for life-threatening acute airway obstruction after removal of the tube. Tracheal narrowing is associated with calcified mediastinal and hilar masses in fibrosing mediastinitis. Tracheal widening has been reported in 30% of patients with pulmonary fibrosis. Primary tracheal neoplasms are rare. In adults 90% of such lesions are malignant, but in children 90% are benign; these benign lesions include squamous cell papilloma and hemangioma. Malignant involvement of the trachea is usually secondary to invasion from adjacent lung, larynx, esophagus or thyroid tissue. Because misplaced tubes are associated with several complications, the radiologist must confirm the location of all tubes. Displacement of the endotracheal tube from full extension to flexion averages 2 cm. Therefore, the tube's tip should be at least this far above the carina to avoid insertion of the tube into the mainstem bronchus. Malpositioning of feeding tubes may cause pneumothorax. Tracheal trauma may be missed unless the radiologist is highly suspicious, as would occur for patients with unresponsive pneumothorax. Observation of the hyoid bone above the third cervical vertebra suggests transection of the cervical trachea.

The trachea: normal anatomic features, imaging and causes of displacement.

Radiologists have long regarded the trachea as one of their "blind spots," even though this structure can be visualized as well as, if not better than, any other intrathoracic component because of the excellent intrinsic contrast provided by air within the trachea and by the tracheal interfaces with the mediastinum and the lung. Despite this ease of visualization, radiologists may overlook tracheal abnormalities in their diligence in examining the pulmonary parenchyma and the heart. The trachea can be imaged by a variety of techniques, including plain radiography, computed tomography (CT) and magnetic resonance imaging. Important tracheal interfaces include the right and left paratracheal stripes and the tracheoesophageal stripe. The right posterolateral tracheal band is best seen with CT. The trachea is generally a midline structure displaced slightly to the right by the aortic arch. Various conditions, including mediastinal masses and vascular anomalies, may bow, displace or indent the trachea. Such appearances are most commonly seen in patients with thyroid masses or a right-sided aortic arch. Enlarged nodes do not usually narrow the trachea unless they are much harder than the cartilaginous rings, as occurs in nodular sclerosing Hodgkin's disease, or the rings are soft, as is the case in children.

Effect of clinical history on the interpretation of chest radiographs in childhood bronchiolitis.

RATIONALE AND OBJECTIVES: This study assesses the effect of a clinical history of "bronchiolitis" versus "well child" on the interpretation of pediatric chest radiographs. METHODS: Radiologists reviewed radiographs previously identified as "definite bronchiolitis," "equivocal bronchiolitis," "definite normal," and "equivocal normal." They were asked to identify features of bronchiolitis and give a confidence level for their diagnosis. A correct clinical history of "bronchiolitis" or a "well child" was provided for all definite and half the equivocal radiographs. The other half of the equivocal radiographs were matched with incorrect histories. RESULTS: Given normal radiographs with a history of bronchiolitis, radiologists reported more features of bronchiolitis with increased confidence and also more features. CONCLUSIONS: The results suggest that the effect of history is not simply to provide additional confirmatory information, thereby increasing the diagnosis rating, but rather to change the interpretation of features, to the extent that features apparently are seen even on normal radiographs.

Research in the clinical radiology department.

Researchers often feel tangled in a web of bureaucracy when attempting to conduct their research. However, it is vital that researchers get involved and develop review processes such as those described above, which facilitate animal research yet address the important ethical, legal, and other related issues raised by hospital administrators and the public. As we do so, it is vital that we communicate directly with the public. Without this participation, we will find more and more hospital radiology departments closing their doors to animal-based research.

The correlation of feature identification and category judgments in diagnostic radiology.

Expert and novice radiologists were given films accompanied by clinical histories that supported a diagnosis either of bronchiolitis or of normal. To provide a plausible task context, some films were radiologically unambiguous and were accompanied by histories consistent with them. For a set of radiologically difficult films from confirmed normal or bronchiolitis patients, fictitious normal or abnormal histories were counterbalanced with the films. The clinical histories affected ratings both of diagnosis and of features present on the difficult films. Thus, uncertainty about individual features evidently was affected by history, and features did not act as an independent source of information. The dependence of feature calls on an overall judgment was also suggested by intra-observer agreement in another study in which an explicit diagnosis was not requested. It is unclear whether the history increased discrimination between normal and abnormal films, or indiscriminately added evidence for or against the disease. Factors are discussed that make it appropriate for feature identification to be partially dependent on category identification.

Intra-aortic balloon pumps: an update on radiographic recognition.

Intra-aortic balloon pumps increase myocardial perfusion and decrease cardiac afterload in patients with cardiogenic shock and in high-risk patients scheduled for cardiac surgery. Appropriate positioning of the pump can be confirmed in chest radiographs. Familiarity with available catheters helps the physician to recognize overadvancement of the balloon in the ascending aorta or the aortic arch vessels. Because the balloon is long, its position is juxtarenal in most patients.

Observer variation in detecting the radiologic features associated with bronchiolitis.

Chest radiographs are commonly obtained to assess children for bronchiolitis, both to corroborate the diagnosis and to exclude other diagnostic possibilities. Their utility in this setting has not previously been examined. Using a blinded, randomized study design, we examined the interobserver and intraobserver variation in the detection of the radiologic features of bronchiolitis from the chest radiograph using "weighted kappa" statistics. This observer variation was compared with that found by other authors for other diagnoses. We also determined the reported presence of these radiologic features in radiographs from patients with bronchiolitis as compared with normal controls. Our study showed acceptable interobserver (kappa = 0.40-0.66) and intraobserver agreement (kappa = 0.50-0.78) on the radiologic features of bronchiolitis relative to other diagnoses. We demonstrated a higher reported presence of these accepted radiologic features in patients with bronchiolitis as compared to controls. Although kappa statistics are widely used in studies of observer variation, "weighted kappa" has received little attention in the radiologic literature. This statistical analysis allows observers to equivocate on the presence or absence of a feature and therefore allows the format of observer variation studies to simulate more closely the normal clinical setting.

Graft pseudoaneurysm simulating desmoid tumor recurrence.

A desmoid tumor is a rare neoplasm of soft tissues and, although benign, is locally invasive and tends to recur. We report a case of suspected recurrence of a supraclavicular desmoid tumor that had been resected 2 years earlier. Computed tomography suggested the correct diagnosis of a pseudoaneurysm of a venous graft that had been placed during the resection.

MR imaging of stump neuroma.

Painful stump neuromas in the postamputation patient are difficult to diagnose and treat. We report a case in which magnetic resonance allowed precise preoperative diagnosis and localization of such a lesion. Magnetic resonance, with its ability for long axis imaging in the extremity, is ideally suited for the workup of stump neuromas.

Sarcoidosis: correlation of pulmonary parenchymal pattern at CT with results of pulmonary function tests.

The appearances of the lungs on radiographs and computed tomographic (CT) scans were correlated with degree of uptake on gallium scans and results of pulmonary function tests (PFTs) in 27 patients with sarcoidosis. CT scans were evaluated both qualitatively and quantitatively. Patients were divided into five categories on the basis of the pattern of abnormality at CT: 1 = normal (n = 4); 2 = segmental air-space disease (n = 4); 3 = spherical (alveolar) masslike opacities (n = 4); 4 = multiple, discrete, small nodules (n = 6); and 5 = distortion of parenchymal structures (fibrotic end-stage sarcoidosis) (n = 9). The percentage of the volume judged to be abnormal (CT grade) was correlated with PFT results for each CT and radiographic category. CT grades were also correlated with gallium scanning results and percentage of lymphocytes recovered from bronchoalveolar lavage (BAL). Patients in CT categories 1 and 2 had normal lung function, those in category 3 had mild functional impairment, and those in categories 4 and 5 showed moderate to severe dysfunction. The overall CT grade correlated well with PFT results expressed as a percentage of the predicted value. In five patients, CT scans showed extensive parenchymal disease not seen on radiographs. CT grades did not correlate with the results of gallium scanning or BAL lymphocytes. The authors conclude that patterns of parenchymal sarcoidosis seen at CT correlate with the PFT results and can be used to indicate respiratory impairment.

Chronic lung diseases: specific diagnosis by using CT.

We evaluated patterns of abnormal lung parenchyma on CT scans in six specific chronic lung diseases and then applied those findings in the differential diagnosis of these lung parenchymal patterns in 56 subjects. There were 48 patients with chronic lung diseases (43 with histologic proof) consisting of usual interstitial pneumonia (n = 20), sarcoidosis (n = 16), lymphangitic carcinomatosis (n = 7), lymphangioleiomyomatosis (n = 2), drug toxicity (n = 2), and eosinophilic granuloma (n = 1). Including eight CT scans of normal control subjects, 56 CT scans were assessed independently by two readers (R1 and R2). Chest radiographs, most of which were obtained within 1 week of CT examination, were available in 48 of the 56 subjects. CT scans were evaluated for specific parenchymal features including disease distribution, lung distortion, thickening of bronchovascular bundles and polygon walls, bronchiectasis, cysts, and nodules, to determine the association of each abnormal feature with the different diseases. Diagnosis was then made from the overall CT appearance of the lungs and, on a separate occasion, from the appearance of the chest radiograph. The correct diagnosis was made from the CT appearance in 54 of 56 patients (R1) and in 50 of 56 patients (R2). Correct diagnoses were made from the chest radiographs in 42 of 48 patients (R1) and 43 of 48 patients (R2). We have identified features that are reproducible and useful when describing CT scans of patients with chronic lung diseases. Interpretation of the appearance of the lung on CT scans was accurate in diagnosing usual interstitial pneumonia, sarcoidosis, and lymphangitic carcinomatosis.

Hickman nodule: a mimic of metastatic disease.

A characteristic small subcutaneous soft-tissue mass can be seen on the computed tomographic scans of patients after removal of Hickman catheters. The mass is indistinguishable from subcutaneous metastatic deposits seen with a variety of malignant tumors. This potentially confusing finding can be differentiated from true metastatic disease by the characteristic location of the subcutaneous nodule in the second or third anterior intercostal space, along the midclavicular line, and by the absence of other subcutaneous masses.

Stump neuroma masquerading as recurrent malignancy on chest roentgenogram.

An amputation neuroma is a nerve's attempt at repair following amputation. While many are asymptomatic, some present as a painful mass. We describe a patient who, following a forequarter amputation for sarcoma, developed a stump neuroma that was unusual because it was first discovered on a routine chest roentgenogram.