Computational reader design and statistical performance evaluation of an in-silico imaging clinical trial comparing digital breast tomosynthesis with full-field digital mammography

Abstract

A recent study reported on an in-silico imaging trial that evaluated the performance of digital breast tomosynthesis (DBT) as a replacement for full-field digital mammography (FFDM) for breast cancer screening. In this in-silico trial, the whole imaging chain was simulated, including the breast phantom generation, the x-ray transport process, and computational readers for image interpretation. We focus on the design and performance characteristics of the computational reader in the above-mentioned trial. Location-known lesion (spiculated mass and clustered microcalcifications) detection tasks were used to evaluate the imaging system performance. The computational readers were designed based on the mechanism of a channelized Hotelling observer (CHO), and the reader models were selected to trend human performance. Parameters were tuned to ensure stable lesion detectability. A convolutional CHO that can adapt a round channel function to irregular lesion shapes was compared with the original CHO and was found to be suitable for detecting clustered microcalcifications but was less optimal in detecting spiculated masses. A three-dimensional CHO that operated on the multiple slices was compared with a two-dimensional (2-D) CHO that operated on three versions of 2-D slabs converted from the multiple slices and was found to be optimal in detecting lesions in DBT. Multireader multicase reader output analysis was used to analyze the performance difference between FFDM and DBT for various breast and lesion types. The results showed that DBT was more beneficial in detecting masses than detecting clustered microcalcifications compared with FFDM, consistent with the finding in a clinical imaging trial. Statistical uncertainty smaller than 0.01 standard error for the estimated performance differences was achieved with a dataset containing approximately 3000 breast phantoms. The computational reader design methodology presented provides evidence that model observers can be useful in-silico tools for supporting the performance comparison of breast imaging systems.

Keywords: channelized hoteling model observer; computational reader; digital breast tomosynthesis; model observer; task-based performance assessment; virtual clinical trial.

Fig. 1

Lesion models of (a) spiculated mass and (b) clustered microcalcifications.

Fig. 2

Sample (a) FFDM images and (b) DBT center slices of the four breast types: dense, heterogeneous, scattered, and fatty, from top to bottom.

Fig. 3

Illustration of original LG channels and the convolutional LG channels for two types of signals: (a) mass and (b) clustered microcalcifications.

Fig. 4

A flowchart of convolutional CHO process illustrated with three channels. In this flowchart, $s^{(1)}$, $s^{(2)}$, and $s^{(3)}$ represent the feature maps of the signal created by convolving signal with the channel functions, and $K_{\lambda }$ represents the covariance of the channel response vector $\lambda ={[{\lambda }_1,{\lambda }_2,{\lambda }_3]}^{\prime }$.

Fig. 5

AUC as a function of LG channel width for detecting the clustered microcalcifications in FFDM images. The optimal channel width is selected to be the point that provides the peak AUC.

Fig. 6

Antler plots (resubstitution and independent testing AUC curves as a function of training sample size) for (a) detecting spiculated mass in FFDM images of heterogeneous breast and (b) detecting clustered microcalcifications in FFDM images of scattered breast.

Fig. 7

Mean signal images of the spiculated mass obtained by averaging the differences between 100 pairs of LP and LA FFDM ROIs of the four breast types: dense, heterogeneous, scattered, and fatty, from right to left.

Fig. 8

Bar plot of $\mathrm{\Delta }\mathrm{AUC}$ (DBT-FFDM) for (a) detecting spiculated mass and (b) detecting clustered microcalcifications in the four breast types.

Fig. 9

The curve of $\mathrm{\Delta }\mathrm{AUC}$ (DBT-FFDM) as a function of the mean AUC of FFDM in the eight tasks. The shaded area ($y>1-x$) indicates the region of impossible $\mathrm{\Delta }\mathrm{AUC}$ values due to a maximum AUC value of 1.