PET/MR Radiomics for Breast Cancer Diagnosis

Breast Cancer

Breast cancer, radiomics, hybrid PET/MRI

Breast cancer is the most common malignancy in women in our country (2013 cancer registry report, Health Promotion Administration). MRI is a more accurate imaging modality for breast lesion diagnosis, monitoring of treatment response, and local staging than compared with mammography and ultrasound. ¹⁸ F-FDG PET was reported to be used for breast cancer diagnosis, staging, and prediction of treatment response as well. We usually interpret the aforementioned imaging modalities by qualitative methods for decision-making. Radiomics is a process involving the conversion of images to quantitative data for subsequent data mining to improve decisional making for patient care, to adjust the patient management, that is so-called precision medicine. Our study is to use semantic and agnostic features of radiomics by hybrid PET/MR for 1. The pre-operative breast cancer patients (without neoadjuvant chemotherapy before operation). 2. The patients will receive neoadjuvant chemotherapy (NAC). The study intends to investigate the association of PET/MR radiomics data with the probability of metastasis or risk of recurrences and survival. We will also investigate if the BD and BPE (measured on MRI) are associated with molecular subtypes, histologic grade and clinical outcome, risk of metastases, and long-term survival of breast cancer patients for the study participants.

Purposes and background introduction: Breast cancer is the most common malignancy in women in our country (2013 cancer registry report, Health Promotion Administration). MRI is a more accurate imaging modality for breast lesion diagnosis, monitoring of treatment response, and local staging than compared with mammography and ultrasound. ¹⁸ F-FDG PET was reported to be used for breast cancer diagnosis, staging, and prediction of treatment response as well. We usually interpret the aforementioned imaging modalities by qualitative methods for decision-making. In recent years, the concept of Radiomics is emerging. Radiomics is a process involving the conversion of images (imaging phenotypes) to quantitative data for subsequent data mining to improve decisional making for patient care, to adjust the patient management, that is so-called precision medicine. Radiomics is applied for the diagnostic, prognostic, and predictive purposes of diseases. There are two main approaches to radiomics: First, the semantic approach, which uses the usual radiological lexicon derived from regions of interest; second, the agnostic approach is higher-order, mathematically computed data derived from images instead of the commonly used radiologists' lexicon. MRI and PET can be used in breast radiomics studies. Hybrid PET/MR is a machine that the PET and MRI can be performed on the same table at the same time slot, therefore, the imaging data of MRI and PET can be obtained at the same examination, with less radiation dosage, more reliable lesion mapping than separate examinations of PET/CT and MRI. Material and methods: There is a total of 120 patients would like to be included in the study. Our study is to use semantic and agnostic features of radiomics by hybrid PET/MR for: 1. The pre-operative breast cancer patients (without neoadjuvant chemotherapy before operation): to investigate the association of PET/MR imaging data with molecular subtypes, cell proliferation (Ki-67), tumor aggressiveness (by histologic grade). 2. The patients who will receive neoadjuvant chemotherapy (NAC): PET/MR study will be performed for 3 times: pre-MAC (study 1), PET/MR after 1st dose of NAC (study 2), and PET/MR after 3rd or 4th NAC (study 3). We will investigate the predictive ability of PET/MR imaging data for NAC response. And we will investigate which parameters at which series of examinations are more predictive of the final NAC response. Therefore, we can adjust the NAC regimen as early as possible. 3. The breast cancer patients mentioned above: we will investigate the association of PET/MR radiomics data with the probability of metastasis or risk of recurrences and survival. 4. We will also investigate if the BD and BPE (measured on MRI) are associated with molecular subtypes, histologic grade and clinical outcome, risk of metastases, and long-term survival of breast cancer patients for the study participants. Data analysis: 1. For the breast cancer patients who will go directly to surgical treatment: We will analyze the association of MRI and PET semantic radiomics features (including DCE MRI, DWI/ADC, CEST, MRS and SUVmax, MTV, TLG) and agnostic radiomics ( texture) features with molecular subtypes, proliferation index (Ki-67), histology type and grades, and tumor size, lymph node status. And we will also investigate whether the semantic or agnostic/ texture analysis, or combination of both can be predictive of the factors associated with clinical outcome (that is, molecular subtype, Ki-67, histology type and grade, size, LN status). 2. For the patients who will receive NAC: As stated previously, each participant is designed to undergo three examinations: Study 1- pre-NAC PET/MR; Study 2- first follow-up PET/MR is performed after first dose of NAC; Study 3, second follow-up PET/MR is performed after third or fourth dose of the NAC. The NAC protocol is mainly antracycline-based followed by taxane-based regimen for a total of 6- 8 cycles. 3. For long-term follow-up:The 5-year survival of each study participant will be obtained, and we will investigate the association of all aforementioned PET or MRI-related radiomics parameters with overall survival, disease-free survival by Cox proportional hazards model. Kaplan-Meier analysis for survival curves and comparison of survival by log-rank test will also be estimated. 4. The association of BPE and breast density (BD) with long term outcome: 5. The Statistical analysis will be performed by Stata 13 (Stata Corp., College Station, Texas, USA) and SAS 9.3 (SAS Institute Inc., Cary, NC, USA). A P value \<0.05 will be regarded as statistical significance.

No related papers found yet.

Taiwan