Abstract

Hon-shing Lam
Radiology of Kowloon West Cluster, Hospital Authority, Hong Kong

The reading of mammogram goes through several processes. Image falls on the retina is recognized by the brain through a subconscious process. A conscious process of detailed focal search then follows. The finding is then analyzed and logical deductions made using existing knowledge. This lecture will describe the various steps we use to perceive and interpret the mammographic image.

Light passes through the mammogram and reaches the retina. Perception of boundary is automatic. Recognition of object needs previous experience and knowledge.

An optimal viewing environment is needed for optimal perception of image. The following factors affect image visualization:

Viewing box brightness: Cone photoreceptors are for detail vision, they are concentrated in the fovea. The cones, however, require an incident luminance of at least 50 NIT (candela/m2) to be fully effective. Excessive illumination, by causing light scatter within the retina, will fog the image. Optimal contrast detail is perceived when the light intensity incident on the eye is 50-500 NIT. The recommended minimum brightness level of viewing box in mammography is 3000 cd m-2

Ambient Light: Ambient light degrades contrast. Complete room darkness, however, is not recommended. Complete darkness will cause pupil dilatation. A constricted pupil has sharper vision as well as greater field depth. Optimal room ambient illumination level should be around 50 lux.

Visual acuity and contrast sensitivity: Deterioration of visual acuity usually occurs with aging. Refraction should be checked annually and corrective lens should be worn if necessary. Magnifying glasses, magnifying views, and use of zoom function in digital image, are other ways to enhance viewing of small lesions.

The process of object recognition occurs at a subconscious level. Brain draws data both from retina and memory. Object is recognized when the data from retina matches that of memory. Studies on eye positioning show that radiograph reading involves two phases: a rapid phase of global visual search followed by a slow phase of focal search. During the rapid phase, the global impression initiates image perception by a Gestalt process. Obvious abnormalities are detected by comparing the mammogram with a previously learned concept of normal. The more experienced the observer, the greater the number of abnormalities detected via this method. Abnormalities are flagged for the second step of focal search. A study of the searching pattern of experts and novice shows that both of them detect most true lesions at the initial 25-40 sec rapid phase. This highlights the importance of the global impression in initiating image perception and directing focal search. The same study also shows that experts tend to find lesions earlier in search than novices, tend to have different fixation and dwell patterns, and tend to have much more efficient search strategies than novices.

To improve one’s performance in visual search of lesions, practice in reading mammographic images is important. Good quality feedback is important to assure that one is learning towards the right direction and not by a random process. Mentor guidance, availability of pathology results, and participation in clinical, radiological and pathology meetings will provide good quality feedback. A good viewing strategy will greatly enhance lesion detection. Varying viewing distances and “look again” are standard strategies. A global impression followed by a second look of the review areas will improve detectability of lesions.

The review areas as suggested by Tabar include the upper outer quadrant on the MLO view; the medial half of the breast, best seen on the CC view; the retroglandular clear space on the CC view; and the retroareolar area. The apex of glandular tissue in MLO view is also an important area. Lesion detection is also enhanced with masking techniques and with the use of mammographic goggles. The presence of a second reader (person or computer) would decrease the number of missed lesions.

After detection of a lesion, it is necessary to decide whether a lesion is real & significant. One would search for confirming visual evidences and rejects irrelevant findings. Concepts in localization and knowledge in analyzing lesion characteristics are helpful. If indicated, additional views and ultrasound should be asked for. Relevant hints such as clinical history & previous mammogram are important. Satisfaction of search, lack of knowledge, deficient training, insufficient experience, fatigue, inattention, distraction & non-belief are the common sources of error. Difficult cases include dense breasts, subtle changes, lack of desmoplastic reaction, slow growing lesions & overlapping features between benign and malignant lesions. Training would improve the knowledge of radiologists. This can be integrated in the day to day work or in attending seminars and workshops. Self assessment programmes are also useful.

References:

1. Commentary : In the eyes of the beholder: what we see is not what we get. B J J Abdullah, and K-H Ng, The British Journal of Radiology, 74 (2001), 675–676
2. American College of Radiology. Mammography quality control manuals for radiologists, radiographic technologists and medical physicists. Reston, Va: American College of Radiology, 1990.
3. Visual Acuity Testing of Radiologists-Is It Time? William H. Straub,1 David Gur, and Barbara C. Good. AJR 156:1107-1108, May 1991
4. A Visual Concept Shapes Image Perception. Harold L. Kundel, Calvin F. Nodine. Radiology 146:363-368, February 1983
5. Time Course of Perception and Decision Making During Mammographic Interpretation. Calvin F. Nodine, Claudia Mello-Thoms, Harold L. Kundel, Susan P. Weinstein. AJR:179, October 2002
6. The effect of search time on perception. Christensen EE, Murry RC, Holland K, Reynolds J, Landay MJ, Moore JG.. Radiology 1981 ;138:361-365
7. The Importance of Perception Research in Medical Imaging. Elizabeth A. Krupinski. Radiation Medicine: Vol. 18 No. 6, 329-334 p.p., 2000
8. Radiology’s Achilles’heel: error and variation in the interpretation of the Roentgen image. P J A Robinson, The British Journal of Radiology, 70 (1997), 1085-1098
9. Analysis of Cancers Missed at Screening Mammography. Richard E.Bird, Terry W. Wallace, Bonnie C.Yankaskas, Radiology 1992; 184: 613-617
10. Missed Breast Carcinoma: Pitfalls and Pearls Aneesa S. Majid, MD, Ellen Shaw de Paredes, MD, Richard D., Doherty, MD, Neil R. Sharma, Xavier Salvador. RadioGraphics 2003; 23:881-895
11. The False Negative Mammogram. Phan T Huynh, Amanda M. Jarolimek, Susanne Daye. RadioGraphics1998; 18:1137-l154