Computer simulations of case difficulty in digital breast tomosynthesis using virtual clinical trials

Barufaldi, Bruno; Vent, Trevor L.; Bakić, Predrag R.; Maidment, Andrew D. A.

doi:10.1002/mp.15553

Cited by 10 publications

(21 citation statements)

References 63 publications

(148 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1 Dr. Glick raises several important questions related to our recently published paper. 2 We can address the point regarding the selection of lesion model I trivially. Lesion model I was included in our paper 2 as a cautionary tale, the lesion differs from the others (II-IV) in that it was not anatomically inspired.…”

Section: In Reply To Glickmentioning

confidence: 99%

“…We thank Dr. Glick for his letter 1 . Dr. Glick raises several important questions related to our recently published paper 2 . We can address the point regarding the selection of lesion model I trivially.…”

mentioning

confidence: 99%

“…In summary, in our paper 2 we describe a method of conducting VCTs in which we vary lesion size, shape, and conspicuity to explore differences in the performance of various modalities. We wrote this paper based upon our own cautionary experiences; that is, we have learned the hard way what happens if you fail to calibrate your VCTs.…”

mentioning

confidence: 99%

See 2 more Smart Citations

In reply to Glick

et al. 2022

Self Cite

View full text Add to dashboard Cite

Section: In Reply To Glickmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

In reply to Glick

et al. 2022

Self Cite

View full text Add to dashboard Cite

“…Furthermore, for the ellipsoidal mass model (mass model A), the calibration process defined the attenuation coefficient of the lesion to be substantially lower than real malignant lesions, with a composition of 10% lesion and 90% normal breast tissue (see Eq. 1 from Barufaldi et al 1 …”

mentioning

confidence: 99%

“…The primary concern is that VCTs should only be calibrated to clinical trial data if the VCT task is similar to that used in the clinical trial. In Barufaldi et al., 1 calibration is used to define the appropriate size and composition of the modeled lesions. The problem with this calibration approach is that the VCT is being calibrated to a clinical reader study that is much different than the VCT study.…”

mentioning

confidence: 99%

In Regards to Barufaldi et al. Med Phys 49(4), 2220–2232

Glick

2022

Medical Physics

View full text Add to dashboard Cite

In Regards to Barufaldi et al. Med Phys 49(4), 2220-2232I read with great interest the recently published paper entitled "Computer simulations of case difficulty in digital breast tomosynthesis using virtual clinical trials" by Barufaldi et al. 1 Clinical trials used to assess safety and effectiveness of digital breast tomosynthesis (DBT) systems are expensive, time consuming, and burdensome, and often prohibit timely patient access to new technologies. The use of virtual clinical trials is gaining interest of late and promises to provide an alternative, quicker, and cost-effective approach for assessing diagnostic accuracy of imaging systems.Barufaldi et al. argue that VCTs need to be "calibrated", in this case by adjusting soft tissue lesion properties (size/shape/composition), so that task performance (area under the ROC curve [AUC]) resulting with the VCT equates to that achieved with human performance measured from a previous clinical reader study (in this case, the study published by Rafferty et al. 2 that compares DBT plus digital mammography [DM] to DM alone). They claim that after calibration, the VCT settings can be used to assess performance of a novel study device,thereby showing equivalency or superiority to the predicate technology.There are many problems and assumptions that are made using this calibration approach and need further discussion. The primary concern is that VCTs should only be calibrated to clinical trial data if the VCT task is similar to that used in the clinical trial. In Barufaldi et al., 1 calibration is used to define the appropriate size and composition of the modeled lesions. The problem with this calibration approach is that the VCT is being calibrated to a clinical reader study that is much different than the VCT study. First, the VCT used to model DBT is being calibrated to a clinical study that used the combination of DM and DBT together, not solely DBT. Second, the evaluation task for the VCT is the so-called signal (or lesion) known-exactly (SKE), location-known-exactly (LKE) problem, where the model observer is given a small region-of -interest (ROI) image and the task is to determine whether a signal is present (located at the center of the ROI) or not. The resulting figure of merit of this experiment is lesion detectability for the SKE/LKE task. The evaluation task for radiologists in the Rafferty et al.'s 2 study that is being calibrated is much more complicated. In that study, radiologists observe both the

show abstract

Microcalcification detectability in breast CT images using CNN observers

Lyu,

Abbey,

Hernandez

et al. 2023

Medical Physics

View full text Add to dashboard Cite

BackgroundBreast computed tomography (CT) is an emerging breast imaging modality, and ongoing developments aim to improve breast CT's ability to detect microcalcifications. To understand the effects of different parameters on microcalcification detectability, a virtual clinical trial study was conducted using hybrid images and convolutional neural network (CNN)‐based model observers. Mathematically generated microcalcifications were embedded into breast CT data sets acquired at our institution, and parameters related to calcification size, calcification contrast, cluster diameter, cluster density, and image display method (i.e., single slices, slice averaging, and maximum‐intensity projections) were evaluated for their influence on microcalcification detectability.PurposeTo investigate the individual effects and the interplay of parameters affecting microcalcification detectability in breast CT.MethodsSpherical microcalcifications of varying diameters (0.04, 0.10, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40 mm) and native intensities were computer simulated to portray the partial volume effects of the imaging system. Calcifications were mathematically embedded into 109 patient breast CT volume data sets as individual calcifications or as clusters of calcifications. Six numbers of calcifications (1, 3, 5, 7, 10, 15) distributed within six cluster diameters (1, 3, 5, 6, 8, 10 mm) were simulated to study the effect of cluster density. To study the role of image display method, 2D regions of interest (ROIs) and 3D volumes of interest (VOIs) were generated using single slice extraction, slice averaging, and maximum‐intensity projection (MIP). 2D and 3D CNNs were trained on the ROIs and VOIs, and receiver operating characteristic (ROC) curve analysis was used to evaluate detection performance. The area under the ROC curve (AUC) was used as the primary performance metric.ResultsDetection performance decreased with increasing section thickness, and peak detection performance occurred using the native section thickness (0.2 mm) and MIP display. The MIP display method, despite using a single slice, yielded comparable performance to the native section thickness, which employed 50 slices. Reduction in slices did not sacrifice detection accuracy and provided significant computational advantages over multi‐slice image volumes. Larger cluster diameters resulted in reduced overall detectability, while smaller cluster diameters led to increased detectability. Additionally, we observed that the presence of more calcifications within a cluster improved the overall detectability, while fewer calcifications decreased it.ConclusionsAs breast CT is still a relatively new breast imaging modality, there is an ongoing need to identify optimal imaging protocols. This work demonstrated the utility of MIP presentation for displaying image volumes containing microcalcification clusters. It is likely that human observers may also benefit from viewing MIPs compared to individual slices. The results of this investigation begin to elucidate how model observers interact with microcalcification clusters in a 3D volume, and will be useful for future studies investigating a broader set of parameters related to breast CT.

show abstract

Computer simulations of case difficulty in digital breast tomosynthesis using virtual clinical trials

Cited by 10 publications

References 63 publications

In reply to Glick

In reply to Glick

In Regards to Barufaldi et al. Med Phys 49(4), 2220–2232

Microcalcification detectability in breast CT images using CNN observers

Contact Info

Product

Resources

About