United States

Background: Advances in the field of radiotherapy planning have led to faster and more accurate planning for patients receiving radiation. One source of this advancement is the use of AI programs to assist with marking target structures and organs at risk (OAR) in a process known as contouring, on patient imaging. This is especially important regarding the HN, given the number and complexity of the organs in this region. 

Methods: Two AI programs dubbed V1 and V2 were utilized to generate contours for four OAR in Eclipse (v15.5, Varian Medical Systems, Inc.). Contours were generated from computed tomography (CT) images of three separate patients receiving radiotherapy for HN cancer at a rural tertiary care center. Dice Similarity Coefficients (DSC), a measure of the spatial overlap between image sets, were calculated between contours created in a medical student-physician team (denoted the G contours) and the AI program contours. Each CT slice for the G, V1, and V2 contours was also scored by a separate medical physicist and radiation oncologist team; these scores were termed subjective scores (SS). A score of 100% corresponded to all slices being deemed acceptable for radiotherapy planning for a given OAR.

Results: The average DSC for the V1 and V2 packages compared to the G contours were 64.6% and 61.6%, respectively (via matched pairs two-tailed t-test: mean difference = -2.98, p-value = 0.03). All SS were above 89%, and the overall average OAR SS were 94.3%, 95.4%, and 92.5% (for the G, V1, and V2 contours, respectively). Two-tailed paired Student’s t-tests between each AI program and the G contours yielded a p-value &gt; 0.05 for the V2 program only.

Conclusion: The V1 and V2 auto contouring packages created OAR contours with similar DSC, though the V1 package demonstrated slightly better overlap with the G contours. SS for the V2 program only were not statistically different from the G contours. This implies the V2 package can create clinically useful contours for HN OAR, at the cost of a slightly decreased DSC compared to V1, translating to faster radiotherapy planning, helping both radiotherapy planning teams and patients alike.

AMA Research Challenge 2022

Evaluating artificial intelligence (AI) programs for head and neck (HN) radiotherapy planning

Welcome to the largest national, multi-specialty research event **featuring research from 750 medical students, residents and fellows, and international medical graduates.**


#### Event Details
The 2022 American Medical Association Research Challenge includes a poster symposium hall and a semifinals hall with the top scored research. Log in at any time throughout the event to:
* **View posters and presentations** on a variety of topics and specialties
* **Score posters in the semifinals** to help decide the top five finalists (AMA member exclusive opportunity)
* **Connect with participants** and provide feedback to help advance their research

At the conclusion of this event, the AMA will announce the top five finalists who will advance to the AMA Research Challenge finals, present their research to an elite panel of judges and compete to win a $10,000 grand prize sponsored by Laurel Road.
 

#### Attend Live Forums
**Oct. 20, 6 — 7 p.m. Central**
 **How to plan a career in academic medicine** 
 Learn how to plan a career in academic medicine and get your questions answered. [Access Here](https://underline.io/events/363/lounge)
 
 **Speakers**
![](https://assets.underline.io/uploads/markdown_image/1/image/d5adb133dfc58eb0a6d469e9cedcd5d4.jpg)
Ricardo Correa, MD, EdD
![](https://assets.underline.io/uploads/markdown_image/1/image/5f9437ab87861538bf02b9ca7c0d926e.jpg)
Stuart M. Greenstein, MD
![](https://assets.underline.io/uploads/markdown_image/1/image/05b1b637c99286d558c371f8e8b185ce.jpg)
Stephanie Strohbeen, MD
 
 **Oct. 22, 11 a.m. — Noon Central**
 **How to get published**
 Learn tips to get your research published and get your questions answered. [Access Here](https://underline.io/events/363/lounge)
 
 **Speakers**
![](https://assets.underline.io/uploads/markdown_image/1/image/8ce6235e957d31cb5e6f9097cc6f6972.jpg)
Jessica Li, MD/PhD Candidate
![](https://assets.underline.io/uploads/markdown_image/1/image/9950ef2e0237f54cb3d8280334d42491.jpg)
Charles Lopresto, DO

### Sponsor
AMA would like to thank Laurel Road for sponsoring a $10,000 grand prize for the winner of the AMA Research Challenge. Laurel Road is not involved in the judging or scoring process to select the winner. The winner will be announced on Dec. 7 during the AMA Research Challenge finals. Payment of the grand prize will be directly from Laurel Road. See the AMA Research Challenge Official Rules for terms and conditions.

![](https://assets.underline.io/uploads/markdown_image/1/image/611e331d4a0eba0920f240cfc04b4d77.png)

**There is no registration fee for this event! **

Register to explore research in a variety of topics. AMA members get exclusive access to score posters in the semifinals.

**Please use the link below to register for free if you have not already done so:**

https://www.ama-assn.org/about/events/ama-research-challenge

Registration Is Required

Welcome to the largest national, multi-specialty research event featuring research from 750 medical students, residents and fellows, and international medical graduates.


Background: Advances in the field of radiotherapy planning have led to faster and more accurate planning for patients receiving radiation. One source of this advancement is the use of AI programs to assist with marking target structures and organs at risk (OAR) in a process known as contouring, on patient imaging. This is especially important regarding the HN, given the number and complexity of the organs in this region. 

Methods: Two AI programs dubbed V1 and V2 were utilized to generate contours for four OAR in Eclipse (v15.5, Varian Medical Systems, Inc.). Contours were generated from computed tomography (CT) images of three separate patients receiving radiotherapy for HN cancer at a rural tertiary care center. Dice Similarity Coefficients (DSC), a measure of the spatial overlap between image sets, were calculated between contours created in a medical student-physician team (denoted the G contours) and the AI program contours. Each CT slice for the G, V1, and V2 contours was also scored by a separate medical physicist and radiation oncologist team; these scores were termed subjective scores (SS). A score of 100% corresponded to all slices being deemed acceptable for radiotherapy planning for a given OAR.

Results: The average DSC for the V1 and V2 packages compared to the G contours were 64.6% and 61.6%, respectively (via matched pairs two-tailed t-test: mean difference = -2.98, p-value = 0.03). All SS were above 89%, and the overall average OAR SS were 94.3%, 95.4%, and 92.5% (for the G, V1, and V2 contours, respectively). Two-tailed paired Student’s t-tests between each AI program and the G contours yielded a p-value > 0.05 for the V2 program only.

Conclusion: The V1 and V2 auto contouring packages created OAR contours with similar DSC, though the V1 package demonstrated slightly better overlap with the G contours. SS for the V2 program only were not statistically different from the G contours. This implies the V2 package can create clinically useful contours for HN OAR, at the cost of a slightly decreased DSC compared to V1, translating to faster radiotherapy planning, helping both radiotherapy planning teams and patients alike.

poster

Harnessing Artificial Intelligence For Improved Detection of Brain Metastases

Purpose: This QI study compared the completeness of HRCT radiology reports before and after the implementation of a disease specific structured reporting template for suspected cases of interstitial lung disease (ILD).

Materials and Methods: A pre-post study of radiology reports for HRCT of the thorax at a multicenter health system was performed. Data was collected in 6-month period intervals before (June 2019-November 2019) and after (January 2021-June 2021) the implementation of a disease specific template. The use of the template was voluntary. The primary outcome measure was the completeness of HRCT reports graded on the basis of the documentation of ten quality measures. The secondary outcome measure assessed the differences in reporting frequencies of each quality measure pre- and post-intervention.

Results: 521 HRCT reports before and 557 HRCT reports after the intervention were reviewed. Of the 557 reports, 118 reports (21%) used the implemented structured reporting template. The mean completeness score of pre-intervention group was 9.20 (SD = 1.08) and post-intervention group was 9.36 (SD = 1.03) with a difference of -0.155, 95% CI [-0.2822, -0.0285, p < 0.0001]. After the intervention, two of the ten quality measures improved significantly: presence of honeycombing from 78.3% to 85.1% (p <0.0039) and technique from 90% to 96.6% (p <0.0001). The rest of the quality measures did not show significant improvement.

Discussion: There are benefits to shifting to disease specific structured reporting for HRCT exams of suspected ILD. Further research on how to improve the voluntary uptake of a disease-specific template is needed to help increase the acceptance of structured reporting among radiologists.

Impact of Structured Reporting Template on the Quality of HRCT Radiology Reports for Interstitial Lung Disease

Purpose: We investigated whether deep learning techniques can be used in enhancing ultra-low-dose 18 F-PI-2620 tau PET/MR images, acquired in a healthy aging population as well as those with neurodegenerative diseases, to produce diagnostic quality images. Methods: 44 total participants were recruited for this study and 18 F-PI-2620 tau PET/MR data were simultaneously acquired. A generative adversarial network was trained to enhance ultra- low-dose tau images, which were reconstructed from a random sampling of 1/20 (approximately
5% of original count level) of the original full-dose data; MR images were also utilized as additional input channels. Region-based analyses as well as a reader study were conducted to assess the image quality of the enhanced images compared to their full-dose counterparts. Results: The enhanced tau images showed significant noise reduction compared to the ultra-low-dose images. The regional standard uptake value ratios showed that while there is an underestimation for both image types, especially in regions with higher uptake, when focusing on the healthy but amyloid positive population, this bias was present in the unenhanced ultra- low-dose images. The radiotracer uptake patterns in the enhanced images were read accurately
compared to their full-dose counterparts. Conclusion: The clinical readings of deep learning-enhanced ultra-low-dose tau PET images were consistent with those performed with full-dose imaging, suggesting the possibility to reduce dose and enable more frequent examinations in patients with dementia.

Using Deep Learning Networks to Improve Ultra-low Dose Tau PET/MR Image Quality

Background
Lung nodules identified by Low-Dose Computed Tomography (LDCT) are classified radiologically in the Lung Reporting and Data System (LungRADS). Herein, we elucidate clinical and radiomic factors associated with follow-up LungRADS 4 categorization and lung cancer diagnosis in patients with initial LungRADS 2-3 nodules.

Methods
We reviewed 2847 patients’ charts with LDCT screening between 2015 and 2019, identifying 2174 (76.7%) with LungRADS 2-3 baseline nodules, of which 908 had follow-up screening available (41.8%). We abstracted gender, race, smoking history, body mass index (BMI), history of chronic obstructive pulmonary disease (COPD), emphysema, pneumonia, cancer, and family history of lung cancer. Association with subsequent LungRADS 4 nodules and lung cancer was assessed via Cox proportional hazards regression. Among a subset of 132 patients, we performed radiomic analysis by annotating the largest nodule of each baseline scan and training machine learning models using 86 textural features. Clinical, radiomic and combined models were developed via Lasso regression and random forest analyses using five-fold cross-validation to predict a subsequent LungRADS 4 result or lung cancer diagnosis.

Results
Among patients with baseline LungRADS 2-3 nodules, 23 (1.1%) developed lung cancer. Age(p=0.04), pack-years (p<0.005), family history of lung cancer (p<0.005), COPD (p=0.02) and emphysema (p<0.005) were associated with lung cancer. 39 patients with follow-up screening (4.3%) progressed to LungRADS 4. Emphysema (p<0.005) was associated with this finding. Machine learning models trained with clinical variables yielded an average area-under-receiver operating characteristic curve (AUC) of 0.543 (random forest) and 0.799 (Lasso regression) for subsequent findings of LungRADS 4 and lung cancer. Radiomic analysis of the largest baseline nodule found an AUC of 0.574 (random forest) and 0.728 (Lasso regression) for findings of LungRADS 4 and lung cancer. Fusion of clinical and radiomic features yielded AUCs of 0.591 (random forest) and 0.718 (Lasso regression) for findings of LungRADS 4 and lung cancer. 


Conclusion
Clinical and radiomic factors showed association with lung cancer diagnosis and LungRADS 4 categorization on follow-up screening in patients with baseline LungRADS 2-3 nodules. Assessment of these features may aid to predict which patients with likely benign screening findings may progress to a higher LungRADS category or develop lung cancer.

Analysis of Clinical and Radiologic factors associated with follow up Lung RADS score and lung cancer diagnosis in a low-intermediate risk lung cancer screening population

Background: The development of tricuspid valve regurgitation (TVR) is a common complication of mitral valve disease. Although severe TVR is usually operated on at the same time as left-sided surgery, there is no clear consensus concerning treatment of mild-moderate TVR.

Methods: We searched electronic databases such as PubMed, Cochrane Library, Scopus, and Web of science to identify studies comparing concomitant tricuspid valve repair at the time of mitral valve surgery to mitral valve surgery alone in patients with mild to moderate TVR. Heterogeneity was measured via the I2 statistic. The Mantel-Haenszel method was used to pool study estimates and the restricted maximum-likelihood estimator was used to estimate between- study heterogeneity. Effect sizes are provided in the form of odds ratios (OR) with 95% confidence intervals (CI).

Results: Our analysis included a total of 25 studies (N = 8,980). The concomitant tricuspid repair group had significantly lower all-cause mortality (OR: 0.57; 95% CI: 0.44 to 0.74) and cardiovascular mortality (OR: 0.49; 95% CI: 0.29 to 0.84). However, the concomitant repair group also had higher odds of permanent pacemaker implantation (OR: 2.09; 95% CI: 1.45 to 3.00). Additionally, while there was no significant difference between the two groups in terms of tricuspid valve reinterventions, stroke, acute kidney injury, concomitant tricuspid valve repair showed statistically significant lower odds for hospitalization due to heart failure (OR: 0.55; 95% CI: 0.22 to 0.58).

Conclusion: Concomitant repair of the tricuspid valve in mild to moderate TVR is a promising intervention to reduce both all-cause and cardiovascular mortality in patients undergoing mitral valve surgeries; however, this risk needs to be weighed against the significantly increased risk of pacemaker implantation.

Concomitant tricuspid valve repair for mild or moderate tricuspid regurgitation for patients undergoing mitral valve surgery? A systematic review and meta-analysis for efficacy and safety

Background: Malperfusion syndrome (MPS) presents a great challenge in treating acute type A aortic dissection (ATAAD) and has a high mortality despite a technically successful operation. We aimed to evaluate the long-term outcomes and progression of the distal aorta in acute type A aortic dissection patients with MPS treated with endovascular fenestration/stenting (F/S) followed by open aortic repair.

Methods: From January 1996-March 2021, 927 patients presented with ATAAD. Five-hundred-and-thirty-four patients had DeBakey I dissection with no-MPS and underwent immediate open repair (no-MPS group), while 97 patients with MPS underwent F/S and delayed open aortic repair (MPS group). Data were collected from the STS database, medical record review, survey, and the national death index (Dec 12, 2021). 

Results: Compared to the no-MPS group, the MPS group had more patients with hypertension (89% vs. 73%, p<0.001), and acute renal failure (60% vs. 4.3%, p<0.001. Both groups had similar aortic root and arch procedures. Postoperatively, the MPS group had similar operative mortality (5.2% vs. 7.9%, p=0.35), more new-onset dialysis (22% vs. 7.7%, p<0.001), similar permanent dialysis (4.7% vs. 2.9%, p=0.50), and prolonged ventilation (72% vs. 49%, p<0.001). The growth rate of the aortic arch (0.38 vs 0.35 mm/year, p=0.81) was similar between the MPS and no-MPS groups. The descending thoracic aorta (1.03 vs. 0.68 mm/year) and abdominal aorta growth rate (0.76 vs. 0.59 mm/year) were significantly higher in the MPS group. The cumulative incidence of reoperation over 10 years (16.2% vs. 17.7%, p=0.91) and 15-year survival outcome (50.4% vs 48.4%, p=0.43) were similar between MPS and no-MPS groups.

Conclusions: Endovascular fenestration/stenting followed by open aortic repair was a valid approach for patients with MPS.

Progression of Distal Aorta after Endovascular Fenestration/Stenting in Acute Type A Aortic Dissection

Background: Azadirachta indica, or neem, is a medicinal plant used commonly throughout South Asia. Neem has demonstrated a wide range of biological activities. This project focuses on the anti-cancer abilities of neem. Our laboratory observations support the anti-cancer effects of neem as a single agent and suggests the ability of neem to improve the efficacy of existing anti-cancer drugs in combination therapies.
Methods: Two different prostate cancer cell lines were used to test the anti-cancer effects of neem. PC3 an androgen receptor negative cell line and C4-2B an androgen independent cell line both have high metastatic potential. Each cell was treated with several different conditions including neem alone, enzalutamide alone, abiraterone alone, and docetaxel alone. Each cell line was also administered combination treatments including neem with enzalutamide, neem with abiraterone and neem with docetaxel. For each of the treatments, a clonogenic assay was used to determine ability of single cell to grow into a colony. Comparisons were analyzed between each of the treatments. 
Results: For both PC3 and C4-2B, enzalutamide alone, abiraterone alone, and docetaxel alone resulted in a lesser number of colonies as compared to the no treatment condition. Treatment conditions involving docetaxel consistently resulted in less than 0-5 colonies per quantification of clonogenic assay. Abiraterone only treatment resulted in less cells compared to no treatment. Neem only and enzalutamide only also resulted in lesser number of colonies though not to the extent of abiraterone and docetaxel. Adjunctive conditions including neem with enzalutamide, neem with abiraterone, and neem with docetaxel produced even less colonies than when the prostate cancer cell lines were treated with their preexisting cancer drug component alone.
Conclusion: As expected, the preexisting prostate cancer drugs (enzalutamide, abiraterone, docetaxel) decreased the ability of single PC3 and C4-2B cells to grow into colonies. Neem alone treatment also appeared to decrease the ability of the prostate cancer cells to grow, although, to a lesser extent than some of the preexisting prostate cancer drugs. Neem as an adjunct to the preexisting cancer drugs appeared to improve the efficacy of the drugs’ ability to decrease cell proliferation.

Anticancer Effect of Azadirachta Indica Products on Prostate Cancer

Background
Patients undergoing radical cystectomy are frequently prescribed prolonged courses of opioids for pain management and have high rates of readmission postoperatively. We aim to determine the association between opioid use and complication rate or length of hospital stay.
Methods
185 total patients who underwent radical cystectomy between 2016 and 2021 were identified in IRB-approved retrospective and prospective databases. data was analyzed using the Wilcoxon-Rank-Sums tests, Chi Squared and Fisher's Exact Tests where appropriate. Multivariate logistic regressions were performed to analyze independent factors associated with postop outcomes. 
Results
This study included 185 patients of median age and BMI of 71.0 and 27.0, respectively. There was no statistically significant difference found in age, BMI, prior tobacco use, or incidence of comorbidities between patients who did and did not require readmission. 21.62% were readmitted within 90 days of surgery. Patients who were readmitted were significantly more likely to have received a longer course of opiates than those who did not undergo readmission. When adjusting for independent demographic variables, multivariate regression analysis demonstrated that each individual day of postoperative opioid use was associated with an increase in the odds of major postoperative complication by 26.84% and an increase in the odds of readmission by 22.23%.
Conclusion
Opioid use represents a possible risk factor for postoperative complications and readmission following radical cystectomy. Further studies are needed to demonstrate if certain classifications of complications are avoidable with conservative administration of opiates in the postoperative period. 



Association between Postoperative Opioid Administration and Surgical Outcomes in Patients Undergoing Radical Cystectomy

Introduction: The gold standard treatment for muscle invasive bladder cancer is radical cystectomy (RC). Robotic assisted radical cystectomy (RARC) has the benefits of decreased blood loss, lower transfusion rates, and shorter length of stay (LOS) when compared to open radical cystectomy (ORC). However, both RARC and ORC are complicated by a high rate of perioperative morbidity with a surgical site complications (SSC) rate ranging from 13-23%. In this study, we sought to determine whether the smaller incisions made during the minimally invasive RARC with intracorporeal urinary diversion (ICUD) resulted in a decreased rate of wound related complications when compared to ORC with extracorporeal urinary diversions (ECUD). 
Methods: We retrospectively reviewed our institutional bladder cancer database for patients undergoing a RC and urinary diversion from 2013 to 2020. Patients were stratified by surgical technique as ORC with ECUD vs. RARC with robotic ICUD. SSC was defined as wound infection, dehiscence, and evisceration and categorized by time of occurrence within 30 days and 31-90 days. Perioperative complications were graded by the Clavien-Dindo classification. Chi-squared analysis was performed to compare the correlation between patient demographics, operative approach, and perioperative characteristics with the presence of SSC.
Results: Of the 273 patients, 128 (46.8%) had ORC with ECUD and 145 (53.1%) had RARC with ICUD. Wound complication rates were significantly lower in patients undergoing the robotic vs. open approach for both 30 day (0.7% vs. 25.0%, p<0.001) and 90 day timeframes (1.4% vs. 28.9%, p<0.001) (Table). Robotic approach led to significantly lower rates of dehiscence vs. open approach in both 30 (0.7% vs. 14.8%, p<0.001) and 90 day totals (1.4% vs. 18.0%, p<0.001). The majority (40.3%) of open and robotic groups were 70-79 years of age (37.5% and 42.8%, p=0.189). 83.5% of patients in both arms had an ileal conduit with the remainder being continent diversions (p=0.02). 
Conclusion: Overall, the SSC rate was significantly lower across all wound complication subcategories (wound infection, dehiscence, and evisceration) in RARC with ICUD which may translate to better perioperative outcomes (shorter LOS and lower readmission rate) and improved patient quality of life. Further studies with larger sample sizes are needed to reinforce the findings.

Next from AMA Research Challenge 2022

Harnessing Artificial Intelligence For Improved Detection of Brain Metastases

Similar lecture

Diagnosing Brain and Bone Cancers Using Artificial Intelligence and Imaging