Purpose

To evaluate the performance of a deep learning (DL) algorithm for clinical measurement of right and left ventricular volume and function across cardiac MR images obtained for a range of clinical indications and pathologies.

Materials and methods

A retrospective, Health Insurance Portability and Accountability Act-compliant study was conducted using the first 200 noncongenital clinical cardiac MRI examinations from June 2015 to June 2017 for which volumetry was available. Images were analyzed using commercially available software for automated DL-based and manual contouring of biventricular volumes. Fully automated measurements were compared using Pearson correlations, relative volume errors, and Bland-Altman analyses. Manual, automated, and expert revised contours for 50 MR images were examined by comparing regional Dice coefficients at the base, midventricle, and apex to further analyze the contour quality.

Results

Fully automated and manual left ventricular volumes were strongly correlated for end-systolic volume (ESV: Pearson r = 0.99, P < .001), end-diastolic volume (EDV: r = 0.97, P < .001), and ejection fraction (EF: r = 0.94, P < .001). Right ventricular measurements were also correlated for ESV (r = 0.93, P < .001), EDV (r = 0.92, P < .001), and EF (r = 0.73, P < .001). Visual inspection of segmentation quality showed most errors (73%) occurred at the cardiac base. Mean Dice coefficients between manual, automated, and expert revised contours ranged from 0.92 to 0.95, with greatest variance at the base and apex.

Conclusion

Fully automated ventricular segmentation by the tested algorithm provides contours and ventricular volumes that could be used to aid expert segmentation, but can benefit from expert supervision, particularly to resolve errors at the basal and apical slices. Supplemental material is available for this article. © RSNA, 2020.

Advances in technology have always had the potential and opportunity to shape the practice of medicine, and in no medical specialty has technology been more rapidly embraced and adopted than radiology. Machine learning and deep neural networks promise to transform the practice of medicine, and, in particular, the practice of diagnostic radiology. These technologies are evolving at a rapid pace due to innovations in computational hardware and novel neural network architectures. Several cutting-edge postprocessing analysis applications are actively being developed in the fields of thoracic and cardiovascular imaging, including applications for lesion detection and characterization, lung parenchymal characterization, coronary artery assessment, cardiac volumetry and function, and anatomic localization. Cardiothoracic and cardiovascular imaging lies at the technological forefront of radiology due to a confluence of technical advances. Enhanced equipment has enabled computed tomography and magnetic resonance imaging scanners that can safely capture images that freeze the motion of the heart to exquisitely delineate fine anatomic structures. Computing hardware developments have enabled an explosion in computational capabilities and in data storage. Progress in software and fluid mechanical models is enabling complex 3D and 4D reconstructions to not only visualize and assess the dynamic motion of the heart, but also quantify its blood flow and hemodynamics. And now, innovations in machine learning, particularly in the form of deep neural networks, are enabling us to leverage the increasingly massive data repositories that are prevalent in the field. Here, we discuss developments in machine learning techniques and deep neural networks to highlight their likely role in future radiologic practice, both in and outside of image interpretation and analysis. We discuss the concepts of validation, generalizability, and clinical utility, as they pertain to this and other new technologies, and we reflect upon the opportunities and challenges of bringing these into daily use.

Purpose

Many medical school clerkships have structured curricula; however, most radiation oncology clerkships do not. The Radiation Oncology Education Collaborative Study Group (ROECSG) implemented a curriculum for fourth-year radiation oncology clerkships at 14 institutions. We hypothesized that students completing clerkships with the curriculum would report greater subjective knowledge and comfort to function as a radiation oncology resident compared with students completing clerkships without the curriculum.

Methods

The ROECSG curriculum included three 1-hour lectures and a 1-hour hands-on radiation treatment planning workshop. Applicants to a single radiation oncology residency program in the 2014-2015 academic year were sent an anonymous, validated clerkship experience survey. Students indicated if clerkships were completed at a curriculum site. Likert-type data (1 = not at all, 5 = extremely) are reported as median (interquartile range).

Results

Respondents described 276 clerkship experiences, of which 64 (23.2%) were completed at a curriculum site. Students whose first clerkship was at a curriculum site perceived greater postclerkship confidence in knowledge of radiation biology (3 [3-4] versus 2 [2-3], P < .01), treatment setup/positioning (3 [2-3] versus 2 [2-3], P < .05), treatment planning (3 [2-3] versus 2 [2-3], P < .01), and ability to integrate evidence-based medicine into treatment (4 [2-4] versus 3 [2-4], P < .05). Students who completed any clerkship with the curriculum had greater postclerkship confidence to function as a radiation oncology resident (3 [3-4] versus 3 [2-3], P < .05).

Conclusions

These results support the curriculum's ability to increase student knowledge in radiation oncology, especially in the students' first clerkship. Further, these findings suggest that expanded implementation of such curricula may ensure a rewarding educational experience during radiation oncology clerkships.

Delivering a cohesive oncology curriculum to medical students is challenging due to oncology's multidisciplinary nature, predominantly outpatient clinical setting, and lack of data describing effective approaches to teaching it. We sought to better characterize approaches to oncology education at US medical schools by surveying third and fourth year medical students who serve on their institution's curriculum committee. We received responses from students at 19 schools (15.2% response rate). Key findings included the following: (1) an under-emphasis of cancer in the curriculum relative to other common diseases; (2) imbalanced involvement of different clinical subspecialists as educators; (3) infrequent requirements for students to rotate through non-surgical oncologic clerkships; and (4) students are less confident in their knowledge of cancer treatment compared to basic science/natural history or workup/diagnosis. Based on these findings, we provide several recommendations to achieve robust multidisciplinary curriculum design and implementation that better balances the clinical and classroom aspects of oncology education.

Purpose

Students applying to radiation oncology residency programs complete 1 or more radiation oncology clerkships. This study assesses student experiences and perspectives during radiation oncology clerkships. The impact of didactic components and number of clerkship experiences in relation to confidence in clinical competency and preparation to function as a first-year radiation oncology resident are evaluated.

Methods and materials

An anonymous, Internet-based survey was sent via direct e-mail to all applicants to a single radiation oncology residency program during the 2012-2013 academic year. The survey was composed of 3 main sections including questions regarding baseline demographic information and prior radiation oncology experience, rotation experiences, and ideal clerkship curriculum content.

Results

The survey response rate was 37% (70 of 188). Respondents reported 191 unique clerkship experiences. Of the respondents, 27% (19 of 70) completed at least 1 clerkship with a didactic component geared towards their level of training. Completing a clerkship with a didactic component was significantly associated with a respondent's confidence to function as a first-year radiation oncology resident (Wilcoxon rank-sum P=.03). However, the total number of clerkships completed did not correlate with confidence to pursue radiation oncology as a specialty (Spearman ρ P=.48) or confidence to function as a first year resident (Spearman ρ P=.43).

Conclusions

Based on responses to this survey, rotating students perceive that the majority of radiation oncology clerkships do not have formal didactic curricula. Survey respondents who completed a clerkship with a didactic curriculum reported feeling more prepared to function as a radiation oncology resident. However, completing an increasing number of clerkships does not appear to improve confidence in the decision to pursue radiation oncology as a career or to function as a radiation oncology resident. These results support further development of structured didactic curricula for the radiation oncology clerkship.

Objective:

Stereotactic ablative radiotherapy (SABR) is being increasingly used as a non-invasive treatment for early-stage non-small cell lung cancer (NSCLC). A non-invasive method to estimate treatment outcomes in these patients would be valuable, especially since access to tissue specimens is often difficult in these cases.

Methods:

We developed a method to predict survival following SABR in NSCLC patients using analysis of quantitative image features on pre-treatment CT images. We developed a Cox Lasso model based on two-dimensional Riesz wavelet quantitative texture features on CT scans with the goal of separating patients based on survival.

Results:

The median log-rank p-value for 1000 cross-validations was 0.030. Our model was able to separate patients based upon predicted survival. When we added tumor size into the model, the p-value lost its significance, demonstrating that tumor size is not a key feature in the model but rather decreases significance likely due to the relatively small number of events in the dataset. Furthermore, running the model using Riesz features extracted either from the solid component of the tumor or from the ground glass opacity (GGO) component of the tumor maintained statistical significance. However, the p-value improved when combining features from the solid and the GGO components, demonstrating that there are important data that can be extracted from the entire tumor.

Conclusions:

The model predicting patient survival following SABR in NSCLC may be useful in future studies by enabling prediction of survival-based outcomes using radiomics features in CT images.

Advances in knowledge:

Quantitative image features from NSCLC nodules on CT images have been found to significantly separate patient populations based on overall survival (p = 0.04). In the long term, a non-invasive method to estimate treatment outcomes in patients undergoing SABR would be valuable, especially since access to tissue specimens is often difficult in these cases.

Purpose/objectives

Three-dimensional (3D) printing is recognized as an effective clinical and educational tool in procedurally intensive specialties. However, it has a nascent role in radiation oncology. The goal of this investigation is to clarify the extent to which 3D printing applications are currently being used in radiation oncology through a systematic review of the literature.

Materials/methods

A search protocol was defined according to preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines. Included articles were evaluated using parameters of interest including: year and country of publication, experimental design, sample size for clinical studies, radiation oncology topic, reported outcomes, and implementation barriers or safety concerns.

Results

One hundred and three publications from 2012 to 2019 met inclusion criteria. The most commonly described 3D printing applications included quality assurance phantoms (26%), brachytherapy applicators (20%), bolus (17%), preclinical animal irradiation (10%), compensators (7%), and immobilization devices (5%). Most studies were preclinical feasibility studies (63%), with few clinical investigations such as case reports or series (13%) or cohort studies (11%). The most common applications evaluated within clinical settings included brachytherapy applicators (44%) and bolus (28%). Sample sizes for clinical investigations were small (median 10, range 1-42). A minority of articles described basic or translational research (11%) and workflow or cost evaluation studies (3%). The number of articles increased over time (P < 0.0001). While outcomes were heterogeneous, most studies reported successful implementation of accurate and cost-effective 3D printing methods.

Conclusions

Three-dimensional printing is rapidly growing in radiation oncology and has been implemented effectively in a diverse array of applications. Although the number of 3D printing publications has steadily risen, the majority of current reports are preclinical in nature and the few clinical studies that do exist report on small sample sizes. Further dissemination of ongoing investigations describing the clinical application of developed 3D printing technologies in larger cohorts is warranted.

Objective:

Stereotactic ablative radiotherapy (SABR) is being increasingly used as a non-invasive treatment for early-stage non-small cell lung cancer (NSCLC). A non-invasive method to estimate treatment outcomes in these patients would be valuable, especially since access to tissue specimens is often difficult in these cases.

Methods:

We developed a method to predict survival following SABR in NSCLC patients using analysis of quantitative image features on pre-treatment CT images. We developed a Cox Lasso model based on two-dimensional Riesz wavelet quantitative texture features on CT scans with the goal of separating patients based on survival.

Results:

The median log-rank p-value for 1000 cross-validations was 0.030. Our model was able to separate patients based upon predicted survival. When we added tumor size into the model, the p-value lost its significance, demonstrating that tumor size is not a key feature in the model but rather decreases significance likely due to the relatively small number of events in the dataset. Furthermore, running the model using Riesz features extracted either from the solid component of the tumor or from the ground glass opacity (GGO) component of the tumor maintained statistical significance. However, the p-value improved when combining features from the solid and the GGO components, demonstrating that there are important data that can be extracted from the entire tumor.

Conclusions:

The model predicting patient survival following SABR in NSCLC may be useful in future studies by enabling prediction of survival-based outcomes using radiomics features in CT images.

Advances in knowledge:

Quantitative image features from NSCLC nodules on CT images have been found to significantly separate patient populations based on overall survival (p = 0.04). In the long term, a non-invasive method to estimate treatment outcomes in patients undergoing SABR would be valuable, especially since access to tissue specimens is often difficult in these cases.

Despite increasing numbers of cancer survivors, non-oncology physicians report discomfort and little training regarding oncologic and survivorship care. This pilot study assesses medical student comfort with medical oncology, surgical oncology, radiation oncology, hospice/palliative medicine, and survivorship care. A survey was developed with input from specialists in various fields of oncologic care at a National Cancer Institute-designated comprehensive cancer center. The survey included respondent demographics, reports of experience with oncology, comfort ratings with oncologic care, and five clinical vignettes. Responses were yes/no, multiple choice, Likert scale, or free response. The survey was distributed via email to medical students (MS1-4) at two US medical schools. The 105 respondents were 34 MS1s (32 %), 15 MS2s and MD/PhDs (14 %), 26 MS3s (25 %), and 30 MS4s (29 %). Medical oncology, surgical oncology, and hospice/palliative medicine demonstrated a significant trend for increased comfort from MS1 to MS4, but radiation oncology and survivorship care did not. MS3s and MS4s reported the least experience with survivorship care and radiation oncology. In the clinical vignettes, students performed the worst on the long-term chemotherapy toxicity and hospice/palliative medicine questions. Medical students report learning about components of oncologic care, but lack overall comfort with oncologic care. Medical students also fail to develop an increased self-assessed level of comfort with radiation oncology and survivorship care. These pilot results support development of a formalized multidisciplinary medical school oncology curriculum at these two institutions. An expanded national survey is being developed to confirm these preliminary findings.

Purpose

Because of the COVID-19 pandemic, the Radiation Oncology Education Collaborative Study Group (ROECSG) hosted its annual international symposium using a virtual format in May 2020. This report details the experience of hosting a virtual meeting and presents attendee feedback on the platform.

Methods and materials

The ROECSG symposium was hosted virtually on May 15, 2020. A postsymposium survey was distributed electronically to assess attendee demographics, participation, and experience. Attendee preference and experience were queried using 3-point and 5-point Likert-type scales, respectively. Symplur LLC was used to generate analytics for the conference hashtag (#ROECSG).

Results

The survey was distributed to all 286 registrants, with a response rate of 67% (191 responses). Seventeen nonattendee responses were omitted from this analysis, for a total of 174 included respondents. Eighty-two attendees (47%) were present for the entire symposium. A preference for a virtual symposium was expressed by 78 respondents (45%), whereas 44 (25%) had no preference and 52 (30%) preferred an in-person meeting. A total of 150 respondents (86%) rated the symposium as "extremely" well organized. Respondents who had not attended a prior in-person ROECSG symposium were more likely to prefer the virtual format (P = .03). Seventy-eight respondents (45%) reported a preference for the virtual platform for reviewing scholarly work, and 103 (59%) reported a preference for an in-person platform for networking. On the day of the symposium, #ROECSG had 408 tweets and 432,504 impressions.

Conclusions

The 2020 ROECSG symposium was well received and can serve as a framework for future virtual meetings. Although the virtual setting may facilitate sharing research, networking aspects are more limited. Effort is needed to develop hybrid virtual and in-person meetings that meet the needs of participants in both settings. Social media is a significant avenue for dissemination and discussion of information and may be valuable in the virtual setting.