Harold L. Kundel

An analysis of perceptual and cognitive factors in radiographic interpretation.

Detection of small tumors in chest x-ray films was studied under tachistoscopic viewing conditions designed to simulate single fixations varying in dwell time, found in free search. Two questions were asked: How much dwell time is required to detect a nodule (experiment 1)? How effective are peripheral inputs in the detection of a nodule (experiment 2)? Our findings indicate that a dwell time of 300 ms was sufficient to detect 85% of the nodules when they were viewed directly. Detection accuracy was reduced by one-half when the tumor was located 5° from the axis of gaze. Taken together with data from eye movement experiments, these results provide useful estimates of perceptual and cognitive parameters of visual search.

Performance of a computer system for recording eye fixations using limbus reflection

A computer system was developed to measure and record locus and duration of the eye axis of gaze using a limbus reflection technique. System hardware and operating procedures for on-line data collection and correction are described. Algorithms are presented that separate fixations from movements. Based on these algorithms, system performance was measured in terms of spatial accuracy of eye fixations and temporal accuracy of saccadic movements.

A computer system for processing eye-movement records

A computer system using the LINC-8 computer was developed to process eye-movement records from two 16-mm cine film input sources : the Mackworth stand camera and the Mackworth wide-angle camera. Data reduction was accomplished by projecting the films onto an X-Y plotting table. An operator marked, using a cursor, either the eyespot, in the case of the stand camera, or the pupil and display center, in the case of the wide-angle camera, in order to digitize the locus of the gaze. This information was normalized and stored on magnetic tape as: ordinal fixation number, fixation duration, and fixation coordinates (X and Y). A software library was created to analyze the three eye-movement measures. One program calculates interfixation distance, fixation time, and dispersion of fixations about a point. A second program operates on data matrices to: calculate area covered by a fixation pattern, frequency of redundant fixations in a fixation pattern, and distribution of fixations over features of the visual display. Data are displayed on either scope or the Calcomp plotter.

Thermographic patterns of the female breast and their relationship to carcinoma

Thermographic patterns of the female breast can be divided into 4 groups: avascular, minimal vascular, marked vascular, and mottled. Carcinoma is slightly more frequent in women with a minimal vascular pattern in the 40‐59 year group and in women with a marked vascular pattern over the age of 60.

A laser scanner for the densitometric analysis of radiographs.

An instrument is described which can scan radiographs and collect densitometric readings for subsequent computer analysis. A laser beam is focused onto the film. Raster scan of the beam is provided by two orthogonal mirror galvanometers. The light, attenuated by the radiograph, is gathered by a lens in such a way that it is directed into a small detector regardless of the scan position. Beam position signals and density readings are stored and analyzed in a minicomputer.

Accuracy Of Radiologists' Decision Making

Reliable studies show that under routine conditions radiologists fail to detect approximately 30% of abnormalities present on chest x-ray films. These failures of detection are perceptual in nature and not due to limitations in such physical parameters as resolution or contrast. There are many image processing techniques that are being proposed to help the radiologist. Most of these techniques have been implemented by us in TV process-ing of x-ray films. Each processing technique was evaluated experimentally using radiologists as observers in a test based on concepts originating from signal detection theory. A numerical value for d, an index of merit, was assigned to each technique according to how well it improved the accuracy of the radiologists decision making.

Radiologic Decision Making And Physical Image Properties

The objective of this work is to relate the performance of radiologists as measured by the probability of detecting a nodular abnormality on a chest film to the properties of the nodule and its surrounding structures measured by a microdensitometer from the film. A parameter called conspicuity has been defined. It consists of measurements of the size, shape, contrast, and the edge gradient of the nodule divided by a measurement of the complexity of the surrounding structures that tend to camouflage the nodule. Conspicuity has been measured for simulated nodules and for a series of real nodules that were originally missed in clinical practice but later detected. Correlation between film reader performance and conspicuity was found.