Impact of Transfer Learning Using Local Data on Performance of a Deep Learning Model for Screening Mammography.

Purpose To investigate the issues of generalizability and replication of deep learning models by assessing performance of a screening mammography deep learning system developed at New York University (NYU) on a local Australian dataset. Materials and Methods In this retrospective study, all individuals with biopsy or surgical pathology-proven lesions and age-matched controls were identified from a South Australian public mammography screening program (January 2010 to December 2016). The primary outcome was deep learning system performance-measured with area under the receiver operating characteristic curve (AUC)-in classifying invasive breast cancer or ductal carcinoma in situ (n = 425) versus no malignancy (n = 490) or benign lesions (n = 44). The NYU system, including models without (NYU1) and with (NYU2) heatmaps, was tested in its original form, after training from scratch (without transfer learning), and after retraining with transfer learning. Results The local test set comprised 959 individuals (mean age, 62.5 years ± 8.5 [SD]; all female). The original AUCs for the NYU1 and NYU2 models were 0.83 (95% CI: 0.82, 0.84) and 0.89 (95% CI: 0.88, 0.89), respectively. When NYU1 and NYU2 were applied in their original form to the local test set, the AUCs were 0.76 (95% CI: 0.73, 0.79) and 0.84 (95% CI: 0.82, 0.87), respectively. After local training without transfer learning, the AUCs were 0.66 (95% CI: 0.62, 0.69) and 0.86 (95% CI: 0.84, 0.88). After retraining with transfer learning, the AUCs were 0.82 (95% CI: 0.80, 0.85) and 0.86 (95% CI: 0.84, 0.88). Conclusion A deep learning system developed using a U.S. dataset showed reduced performance when applied "out of the box" to an Australian dataset. Local retraining with transfer learning using available model weights improved model performance. Keywords: Screening Mammography, Convolutional Neural Network (CNN), Deep Learning Algorithms, Breast Cancer Supplemental material is available for this article. © RSNA, 2024 See also commentary by Cadrin-Chênevert in this issue.

Fluorescent Molecular Imaging Can Improve Intraoperative Sentinel Margin Detection in Oral Squamous Cell Carcinoma.

In head and neck cancer, a major limitation of current intraoperative margin analysis is the ability to detect areas most likely to be positive based on specimen palpation, especially for larger specimens where sampling error limits detection of positive margins. This study aims to prospectively examine the clinical value of fluorescent molecular imaging to accurately identify "the sentinel margin," the point on a specimen at which the tumor lies closest to the resected edge in real-time during frozen section analysis. Methods: Eighteen patients with oral squamous cell carcinoma were enrolled into a prospective clinical trial and infused intravenously with 50 mg of panitumumab-IRDye800CW 1-5 d before surgery. Resected specimens were imaged in a closed-field near-infrared optical imaging system in near real-time, and custom-designed software was used to identify locations of highest fluorescence on deep and peripheral margins. The surgeon identified the sentinel margin masked to optical specimen mapping, and then the regions of highest fluorescence were identified and marked for frozen analysis. Final pathology based on specimen reconstruction was used as reference standard. Results: Resected specimens were imaged in the operating room, and fluorescence had a higher interobserver agreement with pathology (Cohen κ value 0.96) than the surgeon (Cohen κ value of 0.82) for the location of the closest margin. Plotting margin distance at the predicted sentinel margin location of each observer versus the actual closest margin distance at pathology demonstrated best correlation between fluorescence and pathology (R2 = 0.98) with surgeon (R2 = 0.75). Conclusion: Fluorescence imaging can improve identification of the sentinel margin in head and neck cancer resections, holding promise for rapid identification of positive margins and improved oncologic outcomes.

A survey of clinicians on the use of artificial intelligence in ophthalmology, dermatology, radiology and radiation oncology.

Artificial intelligence technology has advanced rapidly in recent years and has the potential to improve healthcare outcomes. However, technology uptake will be largely driven by clinicians, and there is a paucity of data regarding the attitude that clinicians have to this new technology. In June-August 2019 we conducted an online survey of fellows and trainees of three specialty colleges (ophthalmology, radiology/radiation oncology, dermatology) in Australia and New Zealand on artificial intelligence. There were 632 complete responses (n = 305, 230, and 97, respectively), equating to a response rate of 20.4%, 5.1%, and 13.2% for the above colleges, respectively. The majority (n = 449, 71.0%) believed artificial intelligence would improve their field of medicine, and that medical workforce needs would be impacted by the technology within the next decade (n = 542, 85.8%). Improved disease screening and streamlining of monotonous tasks were identified as key benefits of artificial intelligence. The divestment of healthcare to technology companies and medical liability implications were the greatest concerns. Education was identified as a priority to prepare clinicians for the implementation of artificial intelligence in healthcare. This survey highlights parallels between the perceptions of different clinician groups in Australia and New Zealand about artificial intelligence in medicine. Artificial intelligence was recognized as valuable technology that will have wide-ranging impacts on healthcare.

Predicting Professional Examination Outcomes: A Case of the Registered Health Information Administration Certification Examination.

PURPOSE: To determine whether differences existed among candidates for the Registered Health Information Administrator (RHIA) certification examination that may have characterized the likelihood of acquiring professional certification upon graduation. METHODS: Records of total of 197 former students were acquired from accredited health information administration education programs located across the United States. RESULTS: Final course grades in coding and introduction to health information administration and professional curriculum grade point average were strongly associated with the RHIA examination score. Subsequently, a 95% approximate prediction interval was computed. CONCLUSION: Certain academic and demographic variables can be used to predict which students may be at risk for unsuccessfully completing the certification examination. A regression model which can accurately predict the student's examination score was developed.

Acute myocardial infarction and race, sex, and insurance types: unequal processes of care.

Cardiovascular disease is the leading cause of death among men and women in the United States. Georgia's death rate from cardiovascular disease is higher than the national rate. Previous studies have suggested that whites and African Americans do not receive the same processes of care for a first episode of acute myocardial infarction, one of many cardiovascular disease pathologies. Patient's insurance type, race, sex, and whether a patient lives in a rural or urban area have also been suggested as explanations for unequal processes of care for the treatment of acute myocardial infarction. Using diagnosis-related group reimbursement data from an academic medical center in Georgia, this study found independent effects for race (p < .001) and payer status (p = .004), after adjusting for fiscal year. Possible causes for these differences, as well as recommendations for further study, are explored.

An analysis of computerization in primary care practices.

To remain profitable, primary care practices, the front-line health care providers, must provide excellent patient care and reduce expenses while providing payers with accurate data. Many primary care practices have turned to computer technology to achieve these goals. This study examined the degree of computerization of primary care providers in the Augusta, Georgia, metropolitan area as well as the level of awareness of the Health Insurance Portability and Accountability Act (HIPAA) by primary care providers and its potential effect on their future computerization plans. The study's findings are presented and discussed as well as a number of recommendations for practice managers.