Volumetric x-ray coherent scatter imaging of cancer in resected breast tissue: a Monte Carlo study using virtual anthropomorphic phantoms

Abstract

Breast cancer patients undergoing surgery often choose to have a breast conserving surgery (BCS) instead of mastectomy for removal of only the breast tumor. If post-surgical analysis such as histological assessment of the resected tumor reveals insufficient healthy tissue margins around the cancerous tumor, the patient must undergo another surgery to remove the missed tumor tissue. Such re-excisions are reported to occur in 20%-70% of BCS patients. A real-time surgical margin assessment technique that is fast and consistently accurate could greatly reduce the number of re-excisions performed in BCS. We describe here a tumor margin assessment method based on x-ray coherent scatter computed tomography (CSCT) imaging and demonstrate its utility in surgical margin assessment using Monte Carlo simulations. A CSCT system was simulated in GEANT4 and used to simulate two virtual anthropomorphic CSCT scans of phantoms resembling surgically resected tissue. The resulting images were volume-rendered and found to distinguish cancerous tumors embedded in complex distributions of adipose and fibroglandular breast tissue (as is expected in the breast). The images exhibited sufficient spatial and spectral (i.e. momentum transfer) resolution to classify the tissue in any given voxel as healthy or cancerous. ROC analysis of the classification accuracy revealed an area under the curve of up to 0.97. These results indicate that coherent scatter imaging is promising as a possible fast and accurate surgical margin assessment technique.

Coherent scatter imaging Monte Carlo simulation

“…Kapadia et al 9 used a coded-aperture x-ray scatter imaging system and a GEANT4 simulation to study coherent scatter diffraction patterns. Similar results were found by Elshemey et al 10 Lakshmanan et al 11,12 described a tumor margin evaluation method based on x-ray coherent scatter computed tomography imaging using a Monte Carlo Geant4 code. The resulting images distinguish cancerous tumors embedded in complex distributions of adipose and fibroglandular tissue.…”

Accuracy assessment and characterization of x-ray coded aperture coherent scatter spectral imaging for breast cancer classification

“…Once the image of the sample in x-z-q space has been acquired, the next step is to determine which regions or pixels in the sample are cancerous. Similar to the technique that we used for identifying cancer for CSCT imaging 38 and the accompanying ROC analysis, here for CACSSI, we analyzed each pixel of the sample in x-z space individually and diagnosed it as cancerous based on its Euclidean proximity in q-space to the expected behavior for cancerous tissue. The expected behavior for cancer was taken from the scatter intensity for cancerous tissue as a function of q that was measured in Ref.…”

Design and implementation of coded aperture coherent scatter spectral imaging of cancerous and healthy breast tissue samples

“…2(a)] taken from Ref. 27 was formed by inserting a mathematical lesionwhose shape is based on tomosynthesis images of breast tumors-into a 25-mm cube containing a distribution of healthy breast tissues taken from a segmented cone beam CT image of a healthy patient's breast. Specifically, the lesion shape and size (∼1.92 cm) were generated using a mathematical tumor model based on lesions segmented from high resolution transmissionbased tomosynthesis images; the mathematical tumor model that was used was previously shown to be indistinguishable by radiologists from lesions in CT images.…”