Logical Analysis in Roentgen Diagnosis

Memorial Fund Lecture1
Published Online:https://doi.org/10.1148/74.2.178

During the first year of my residency training I had occasion to ask my chief, Dr. Robert Stone, how I could be sure of the accuracy of my roentgen diagnosis. In the course of our discussion he told me about his former chief, Dr. Howard Ruggles, an excellent diagnostician who was able to make a high percentage of correct diagnoses after a brief examination of the films. Sometimes physicians felt that Dr. Ruggles' ready diagnoses were perhaps not too reliable and they might ask: “Dr. Ruggles, what makes you think this shadow is a metastatic lesion?” To which his just as ready reply would be: “Because it looks like it!” All of us to some extent make diagnoses in this way, and often it is difficult to describe the many possible variations of certain lesions, as, for instance, the protean manifestations of pulmonary tuberculosis.

But what about the traps set for the unwary radiologist who uses this “looks like it” method too freely? Figure 1 shows the lung manifestations of four different diseases. The roentgen findings are similar and few radiologists would make all four diagnoses correctly on the basis of the film alone. However, if I told you that the first patient had carcinoma of the stomach, the second had pneumococci in the sputum, the third had scleroderma, and the fourth appeared acutely ill with dyspnea, cough, and fever, a white blood count of 10,000, and negative sputum culture, you would be much more likely to give the correct diagnoses, namely, for patient 1, lymphatic spread of carcinoma of the stomach; patient 2, disseminated pneumococcal pneumonia; patient 3, diffuse fibrosis in scleroderma; and patient 4, diffuse interstitial pneumonitis of viral etiology.

Bone lesions are also difficult to describe and identify. Figure 2 is the roentgenogram of the right forearm of a woman who was examined because of her complaint of pain in the arm. An expanding lesion is present in the middle third of the radius. The cortex is intact; there is no periosteal elevation, and a coarse trabecular pattern is present. In this case, if I told you that the patient's left kidney had been removed because of a malignant tumor, you would at once suggest that the bone lesion very likely is a metastatic hypernephroma. Later I will have more to say concerning the description and identification of bone tumors.

It was to problems of identification such as these that Dr. Sosman was referring in his 1950 Shattuck Lecture on The Specificity and Reliability of Roentgenographic Diagnosis (1) when he said, “Our diagnoses are based on gross pathology (that is, disturbed morphology) in the great majority of cases—certainly well over 90 per cent. As roentgenologists I am sure that much of our accuracy depends on mathematical probabilities in a given case or set of circumstances.”

Article History

Published in print: Feb 1960