Multitask deep learning on mammography to predict extensive intraductal component in invasive breast cancer.

OBJECTIVES: To develop a multitask deep learning (DL) algorithm to automatically classify mammography imaging findings and predict the existence of extensive intraductal component (EIC) in invasive breast cancer. METHODS: Mammograms with invasive breast cancers from 2010 to 2019 were downloaded for two radiologists performing image segmentation and imaging findings annotation. Images were randomly split into training, validation, and test datasets. A multitask approach was performed on the EfficientNet-B0 neural network mainly to predict EIC and classify imaging findings. Three more models were trained for comparison, including a single-task model (predicting EIC), a two-task model (predicting EIC and cell receptor status), and a three-task model (combining the abovementioned tasks). Additionally, these models were trained in a subgroup of invasive ductal carcinoma. The DeLong test was used to examine the difference in model performance. RESULTS: This study enrolled 1459 breast cancers on 3076 images. The EIC-positive rate was 29.0%. The three-task model was the best DL model with an area under the curve (AUC) of EIC prediction of 0.758 and 0.775 at the image and breast (patient) levels, respectively. Mass was the most accurately classified imaging finding (AUC = 0.915), followed by calcifications and mass with calcifications (AUC = 0.878 and 0.824, respectively). Cell receptor status prediction was less accurate (AUC = 0.625-0.653). The multitask approach improves the model training compared to the single-task model, but without significant effects. CONCLUSIONS: A mammography-based multitask DL model can perform simultaneous imaging finding classification and EIC prediction. CLINICAL RELEVANCE STATEMENT: The study results demonstrated the potential of deep learning to extract more information from mammography for clinical decision-making. KEY POINTS: • Extensive intraductal component (EIC) is an independent risk factor of local tumor recurrence after breast-conserving surgery. • A mammography-based deep learning model was trained to predict extensive intraductal component close to radiologists' reading. • The developed multitask deep learning model could perform simultaneous imaging finding classification and extensive intraductal component prediction.

Different impacts of cancer types on cancer screening during COVID-19 pandemic in Taiwan.

BACKGROUND: The COVID-19 pandemic has rapidly become a major challenge for global health care systems and affected other priorities such as the utilization of population-based cancer screening services. We sought to examine to what extent the COVID-19 pandemic has affected cancer screening utilization in Taiwan, even the use of inreach and outreach screening services for different types of cancer screening and different regions. METHODS: Using nationwide cervical, breast, colorectal and oral cancer screening data, the percentage changes in screening participants at inreach and outreach services were calculated and compared between January to April 2020 (COVID-19 pandemic) and January to April 2019. RESULTS: The average percentage change declined from 15% to 40% for cervical, breast, and colorectal cancer screening, with a nearly 50% decline in oral cancer screening. There was a greater preference for breast and colorectal cancer screening outreach services, which had greater accessibility and declined less than inreach services in most regions. The screening utilization varied in different regions, especially in eastern Taiwan where the less convenient transportation and lower risk of COVID-19 transmission had a positive change on four types of cancer screening outreach services. CONCLUSION: The COVID-19 pandemic may have had an effect not only in the utilization of different types of cancer screening but also in the preference between inreach and outreach services, and even in variations in screening services in different regions.

Computer-aided detection system for chest radiography: reducing report turnaround times of examinations with abnormalities.

BACKGROUND: The ability to give high priority to examinations with pathological findings could be very useful to radiologists with large work lists who wish to first evaluate the most critical studies. A computer-aided detection (CAD) system for identifying chest examinations with abnormalities has therefore been developed. PURPOSE: To evaluate the effectiveness of a CAD system on report turnaround times of chest examinations with abnormalities. MATERIAL AND METHODS: The CAD system was designed to automatically mark chest examinations with possible abnormalities in the work list of radiologists interpreting chest examinations. The system evaluation was performed in two phases: two radiologists interpreted the chest examinations without CAD in phase 1 and with CAD in phase 2. The time information recorded by the radiology information system was then used to calculate the turnaround times. All chest examinations were reviewed by two other radiologists and were divided into normal and abnormal groups. The turnaround times for the examinations with pathological findings with and without the CAD system assistance were compared. RESULTS: The sensitivity and specificity of the CAD for chest abnormalities were 0.790 and 0.697, respectively, and use of the CAD system decreased the turnaround time for chest examinations with abnormalities by 44%. CONCLUSION: The turnaround times required for radiologists to identify chest examinations with abnormalities could be reduced by using the CAD system. This system could be useful for radiologists with large work lists who wish to first evaluate the most critical studies.

Integration of Clinical and CT-Based Radiomic Features for Pretreatment Prediction of Pathologic Complete Response to Neoadjuvant Systemic Therapy in Breast Cancer.

The purpose of the present study was to examine the potential of a machine learning model with integrated clinical and CT-based radiomics features in predicting pathologic complete response (pCR) to neoadjuvant systemic therapy (NST) in breast cancer. Contrast-enhanced CT was performed in 329 patients with breast tumors (n = 331) before NST. Pyradiomics was used for feature extraction, and 107 features of seven classes were extracted. Feature selection was performed on the basis of the intraclass correlation coefficient (ICC), and six ICC thresholds (0.7−0.95) were examined to identify the feature set resulting in optimal model performance. Clinical factors, such as age, clinical stage, cancer cell type, and cell surface receptors, were used for prediction. We tried six machine learning algorithms, and clinical, radiomics, and clinical−radiomics models were trained for each algorithm. Radiomics and clinical−radiomics models with gray level co-occurrence matrix (GLCM) features only were also built for comparison. The linear support vector machine (SVM) regression model trained with radiomics features of ICC ≥0.85 in combination with clinical factors performed the best (AUC = 0.87). The performance of the clinical and radiomics linear SVM models showed statistically significant difference after correction for multiple comparisons (AUC = 0.69 vs. 0.78; p < 0.001). The AUC of the radiomics model trained with GLCM features was significantly lower than that of the radiomics model trained with all seven classes of radiomics features (AUC = 0.85 vs. 0.87; p = 0.011). Integration of clinical and CT-based radiomics features was helpful in the pretreatment prediction of pCR to NST in breast cancer.

Cost-effectiveness of stereotactic vacuum-assisted biopsy for nonpalpable breast lesions.

PURPOSE: To examine the potential cost-savings of stereotactic vacuum-assisted biopsy (SVAB) over open surgical biopsy (OSB) in diagnosis of nonpalpable lesions on mammography and to estimate the cost-saving effect on lesions at different levels of malignant probability. METHODS: This retrospective study was approved by our Institutional Review Board. We retrospectively reviewed 276 (33.8 %) SVAB and 541 (66.2 %) OSB medical records at a medical center. Direct costs included patients' self-paid and national health insurance claim charges. Indirect costs were calculated using sick days, average salary, and age-adjusted employment rate. One-way and two-way sensitivity analyses were conducted. Lesion classification was determined by the assessment categories of Breast Imaging Reporting and Data System (BI-RADS), 4th or 5th editions. RESULTS: SVAB decreased the direct cost by $90.3 (10.1 %) per diagnosis. The indirect cost was decreased by $560.2 (96.0 %). Overall, SVAB saved 43.9 % of resource utilization for each biopsy. Taking the cost of the subsequent malignant surgery into account, from the healthcare providers' perspective, SVAB was cost-effective if a lesion had less than 19 % likelihood of malignancy. From the societal perspective, SVAB reduced productivity loss for all the lesions. Based on the positive predictive value of the BI-RADS categories, SVAB was more suitable for the lesions of category 4A and category 3, resulting in greater savings in both medical and societal resources. CONCLUSIONS: SVAB is a cost-effective diagnostic option for nonpalpable breast lesions. The cost-saving effect is greater for the lesions of category 4A and category 3.

Effects of the COVID-19 pandemic on breast cancer screening in Taiwan.

The breast cancer screening program has continued in Taiwan during the COVID-19 pandemic. Our nationwide data showed that the total number of screenings decreased by 22.2%, which was more pronounced for in-hospital examinations (-37.2%), while outreach showed a 12.9% decrease. This decline in screening participation happened at all levels of hospitals, more significantly at the highest level. Our report revealed that outreach services could maintain relatively stable breast cancer screening under this kind of public health crisis. Building a flexible, outreach system into the community might need to be considered when policymakers are preparing for future possible pandemics.

Accuracy and outcomes of stereotactic vacuum-assisted breast biopsy for diagnosis and management of nonpalpable breast lesions.

Stereotactic vacuum-assisted biopsy (SVAB) is an alternative method of breast biopsy for nonpalpable lesions detected by mammography. Considering the diagnostic effectiveness, a direct comparison of SVAB and open surgical biopsy (OSB) is lacking. We performed a retrospective review of 276 (33.8%) SVAB and 541 (66.2%) OSB to compare the diagnostic accuracy and the total number of procedures the patients underwent. The negative predictive values of OSB and SVAB were 99.77% and 99.61%, and their false-negative rates were 0.96% and 4.76%, respectively. SVAB, as the first-line biopsy method, obviated 92.3% of operations. All malignancies diagnosed using SVAB could be treated with single therapeutic surgery. By contrast, 48% of malignancies of OSB group received two operations. Breast Imaging Reporting and Data System (BI-RADS) category used at the study correlated well with the percentage of malignancy and can thus be used to predict biopsy results. Our study concluded that SVAB is reliable for diagnosing nonpalpable breast lesions and is the better biopsy method for categories 3 and 4A lesions, which reduces the benign surgery rate. For lesions with a higher likelihood of malignancy, BI-RADS 4B, 4C and 5, SVAB has an advantage over OSB, which lowers the total number of operations for malignancy treatment.

3-T proton magnetic resonance spectroscopy of central neurocytoma: 3 case reports and review of the literature.

Central neurocytoma (CNC), first described by Hassoun et al in 1982, is a rare neuronal tumor of the central nervous system, accounting for 0.25% to 0.5% of all central nervous system tumors. To our knowledge, there are only 5 published articles reporting the magnetic resonance spectroscopy (MRS) findings of neurocytomas. The 3-T proton MRS findings of 3 cases with CNC confirmed by immunohistochemical stains are reported here. Increased choline (Cho)/creatine (Cr) ratios with decreased N-acetylaspartate (NAA)/Cr ratios were observed in all 3 cases, but only 1 case had an increased peak at 3.55 ppm known as glycine (Gly). The other case with an increased alanine peak at 1.5 ppm had a poor prognosis. Therefore, we conclude that the presence of a Gly peak may suggest the diagnosis of CNC but that the absence of Gly does not exclude the diagnosis of CNC.