Addressing the Challenges in Transitioning from Mammography to Breast Tomosynthe*

描述（由申请人提供）：本K25奖将用于物理学家和光学工程师Mini Das博士成功过渡，将她的技能应用于乳腺成像的临床相关研究。她的长期目标是成为该领域的独立研究人员，并开发一个研究项目，将创新物理学与相关生物学过程的透彻理解、最先进的成像技术和早期乳腺癌检测的实际临床需求相结合。在本提案中，Das博士将解决从2D乳房x光检查到3D乳房图像过渡中的一些现有挑战。数字乳房x线摄影（DM）显示微钙化（MCs）的检出率非常高，微钙化是浸润性导管癌（DCIS）的主要前兆。虽然数字乳腺断层合成（DBT）是一种等待FDA批准的模式，但目前尚不清楚它是否足以取代DM。由于MC和肿块之间的衰减特性和尺寸/结构的差异，很难优化DBT系统，以获得最佳性能，从同一重建图像中同时检测MC和肿块。本文提出的可变剂量DBT （VD-DBT）技术评价具有重要意义。在这里，可以同时获得相当于DM的中心投影切片（主要用于MC检测）和3D重建图像（主要用于质量检测）。该方法还将为放射科医生提供一个平滑的过渡，以尽可能低的总剂量同时获得乳房x线摄影图像和3D重建图像。我们提出的第二部分是研究具有可调先验的惩罚最大似然（PML）重建方法，该方法可以实现具有提高信噪比的低噪声图像，并且边缘保持效果可以改善使用3D重建图像的质量检测。考虑到在任何拟议的乳房筛查方式（如DBT）中都需要减少总成像剂量，这一点也非常重要。

DESCRIPTION (provided by applicant): This K25 award will be used for successful transition for Dr. Mini Das, who is a physicist and optical engineer, to apply her skills to clinically relevant research in breast imaging. Her long-term goals are to be an independent researcher in this area and to develop a research program combining innovative physics with a thorough understanding of relevant biological processes, state of the art imaging techniques and real clinical needs for early stage breast cancer detection. In this proposal, Dr. Das will address some of the existing challenges in transitioning from 2D mammography to 3D breast images. Digital mammography (DM) has shown very high rate of detectability for microcalcifications (MCs), which are the primary precursor for invasive ductal carcinoma (DCIS). While digital breast tomosynthesis (DBT) is a modality waiting for FDA approval, it is not clear if it has been proven good enough to replace DM. Due to the differences in the attenuation properties and size/structure between MCs and masses, it is difficult to optimize DBT systems for best performance for both MC and mass detection from the same reconstructed image In light of these various challenges, the evaluation of the proposed variable dose DBT (VD-DBT) technique proposed here is of high significance. Here, a central projection slice equivalent to DM (primarily for MC detection) and a 3D reconstruction image (primarily for mass detection) can be acquired simultaneously. This method will also provide a smooth transition for the radiologists to obtain both a mammographic image and a 3D reconstructed image simultaneously with the lowest possible total dose. The second part of our proposal to investigate a penalised maximum likelihood (PML) reconstruction method with adjustable priors that can enable low noise image with improved SNR and edge preserving effects can improve the detection of masses using the 3D reconstructed image. This is also highly significant considering the need to reduce the total imaging dose in any proposed breast screening modality like DBT.