METHODS: We performed a literature search of published articles on the application of MRI phenotypic features in invasive breast cancer molecular subtype classifications by radiologists' interpretation on Medline Complete, Pubmed, and Google scholar from 1st January 2000 to 31st March 2021. Of the 1453 literature identified, 42 fulfilled the inclusion criteria.
RESULTS: All studies were case-controlled, retrospective study and research-based. The majority of the studies assessed the MRI features using American College of Radiology- Breast Imaging Reporting and Data System (ACR-BIRADS) classification and using dynamic contrast-enhanced (DCE) kinetic features, Apparent Diffusion Coefficient (ADC) values, and T2 sequence. Most studies divided invasive breast cancer into 4 main subtypes, luminal A, luminal B, HER2, and triple-negative (TN) cancers, and used 2 readers. We present a summary of the radiologists' extracted breast MRI phenotypical features and their correlating breast cancer subtypes classifications. The characteristic features are morphology, enhancement kinetics, and T2 signal intensity. We found that the TN subtype has the most distinctive MRI features compared to the other subtypes and luminal A and B have many similar features.
CONCLUSION: The MRI features which are predictive of each subtype are the morphology, internal enhancement features, and T2 signal intensity, predominantly between TN and the rest. Radiologists' visual interpretation of some of MRI features may offer insight into the respective invasive breast cancer molecular subtype. However, current evidence are still limited to "suggestive" features instead of a diagnostic standard. Further research is recommended to explore this potential application, for example, by augmentation of radiologists' visual interpretation by artificial intelligence.
METHODS: In this article, the steps involved in importing, segmenting, and registering tomographic images using 3D Slicer were thoroughly described, before importing them into GATE for MC simulation. The absorbed doses estimated using GATE were then compared with that of PM. SlicerRT, a 3D Slicer extension, was then used to visualize the isodose from the MC simulation.
RESULTS: A workflow diagram consisting of all the steps taken in the utilization of 3D Slicer for personalized dosimetry in 90 Y radioembolization has been presented in this article. In comparison to the MC simulation, the absorbed doses to the tumor and normal liver were overestimated by PM by 105.55% and 20.23%, respectively, whereas for lungs, the absorbed dose estimated by PM was underestimated by 25.32%. These values were supported by the isodose distribution obtained via SlicerRT, suggesting the presence of beta particles outside the volumes of interest. These findings demonstrate the importance of personalized dosimetry for a more accurate absorbed dose estimation compared to PM.
CONCLUSION: The methodology provided in this study can assist users (especially students or researchers who are new to MC simulation) in navigating intricate steps required in the importation of tomographic data for MC simulation. These steps can also be utilized for other radiation therapy related applications, not necessarily limited to internal dosimetry.
METHODS: We propose to use Residual Blocks with a 3 × 3 kernel size for local feature extraction and Non-Local Blocks to extract the global features. The Non-Local Block has the ability to extract global features without using a huge number of parameters. The key idea behind the Non-Local Block is to apply matrix multiplications between features on the same feature maps.
RESULTS: We trained and validated the proposed method on the LIDC-IDRI dataset which contains 1018 computed tomography scans. We followed a rigorous procedure for experimental setup, namely tenfold cross-validation, and ignored the nodules that had been annotated by