Automated segmentation of phases, steps, and tasks in laparoscopic cholecystectomy using deep learning.

BACKGROUND: Video-based review is paramount for operative performance assessment but can be laborious when performed manually. Hierarchical Task Analysis (HTA) is a well-known method that divides any procedure into phases, steps, and tasks. HTA requires large datasets of videos with consistent definitions at each level. Our aim was to develop an AI model for automated segmentation of phases, steps, and tasks for laparoscopic cholecystectomy videos using a standardized HTA. METHODS: A total of 160 laparoscopic cholecystectomy videos were collected from a publicly available dataset known as cholec80 and from our own institution. All videos were annotated for the beginning and ending of a predefined set of phases, steps, and tasks. Deep learning models were then separately developed and trained for the three levels using a 3D Convolutional Neural Network architecture. RESULTS: Four phases, eight steps, and nineteen tasks were defined through expert consensus. The training set for our deep learning models contained 100 videos with an additional 20 videos for hyperparameter optimization and tuning. The remaining 40 videos were used for testing the performance. The overall accuracy for phases, steps, and tasks were 0.90, 0.81, and 0.65 with the average F1 score of 0.86, 0.76 and 0.48 respectively. Control of bleeding and bile spillage tasks were most variable in definition, operative management, and clinical relevance. CONCLUSION: The use of hierarchical task analysis for surgical video analysis has numerous applications in AI-based automated systems. Our results show that our tiered method of task analysis can successfully be used to train a DL model.

A Novel Needle Mouse Model of Vascular Cognitive Impairment and Dementia.

Bilateral common carotid artery (CCA) stenosis (BCAS) is a useful model to mimic vascular cognitive impairment and dementia (VCID). However, current BCAS models have the disadvantages of high cost and incompatibility with magnetic resonance imaging (MRI) scanning because of metal implantation. We have established a new low-cost VCID model that better mimics human VCID and is compatible with live-animal MRI. The right and the left CCAs were temporarily ligated to 32- and 34-gauge needles with three ligations, respectively. After needle removal, CCA blood flow, cerebral blood flow, white matter injury (WMI) and cognitive function were measured. In male mice, needle removal led to ∼49.8% and ∼28.2% blood flow recovery in the right and left CCA, respectively. This model caused persistent and long-term cerebral hypoperfusion in both hemispheres (more severe in the left hemisphere), and WMI and cognitive dysfunction in ∼90% of mice, which is more reliable compared with other models. Importantly, these pathologic changes and cognitive impairments lasted for up to 24 weeks after surgery. The survival rate over 24 weeks was 81.6%. Female mice showed similar cognitive dysfunction, but a higher survival rate (91.6%) and relatively milder white matter injury. A novel, low-cost VCID model compatible with live-animal MRI with long-term outcomes was established.SIGNIFICANCE STATEMENT Bilateral common carotid artery (CCA) stenosis (BCAS) is an animal model mimicking carotid artery stenosis to study vascular cognitive impairment and dementia (VCID). However, current BCAS models have the disadvantages of high cost and incompatibility with magnetic resonance imaging (MRI) scanning due to metal implantation. We established a new asymmetric BCAS model by ligating the CCA to various needle gauges followed by an immediate needle removal. Needle removal led to moderate stenosis in the right CCA and severe stenosis in the left CCA. This needle model replicates the hallmarks of VCID well in ∼90% of mice, which is more reliable compared with other models, has ultra-low cost, and is compatible with MRI scanning in live animals. It will provide a new valuable tool and offer new insights for VCID research.

Partnership Conference.

BACKGROUND: The Duke Global Health Institute (DGHI) was founded in 2006 with a goal to foster interdisciplinary global health education and research across Duke University and Duke Medical Center. Critical to achieving this goal is the need to develop and sustain strong international partnerships. OBJECTIVE: To host a conference with multiple international partners and strengthen existing relationships. METHODS: After a deliberate year-long planning process, DGHI convened a Partnership Conference with its international partners on the Duke University campus in conjunction with its 10th Anniversary Celebration. The Partnership Conference sought to promote an exchange of novel ideas in support of global health education and research, explore new collaborations in South-South relationships, and identify and facilitate pursuit of new educational and research opportunities. FINDINGS: A total of 25 partners from 10 countries and 46 DGHI faculty members participated in the 3-day event in October 2016. Activities included workshops on preselected research topics, educational symposia on novel teaching methods and harnessing technological advances, introduction of the Health Humanities Laboratory to prepare students and trainees for fieldwork, and discussions of research infrastructure and training needs. Surveys from visiting partners revealed a high degree of satisfaction. Proposed action items include methods to realize improved communications, enhancement of mutual education opportunities, support and mentoring to build local research capacity, and more exchange of faculty and students between partnering institutions. CONCLUSIONS: With careful planning from all parties, a multilateral partnership conference including both university and medical center faculty can be a productive forum for exchange on global health education and research. Sustaining such partnerships is vital to the success of global health scholarship.

Critical Evaluation of Delayed Healing of Venous Leg Ulcers: A Retrospective Analysis in Canadian Patients.

BACKGROUND: Venous leg ulcers (VLUs) are common but challenging health problems. Better understanding of the risk factors involved in delayed healing of VLUs may therefore guide individualized treatment plans to improve patient outcomes. OBJECTIVE: We sought to identify the risk factors associated with delayed healing of VLUs in patients seen at a tertiary academic wound care clinic. METHODS: A retrospective analysis of 554 patients who presented to the Toronto Regional Wound Healing Clinic for VLUs in a 3-year period was performed. Patient and ulcer characteristics were recorded. Multivariate analyses were performed to compare patients with ulcer resolution and those whose ulcers did not resolve after 12-month follow-up. RESULTS: The average age of the patients was 67.3 ± 0.7 years, with 56 % being female. The most common comorbidities were hypertension (54 %), dyslipidemia (33 %), a history of smoking (30 %), and diabetes (26 %). Ulcer resolution was associated with a smaller ulcer size (odds ratio [OR] 0.984 [95 % confidence interval (CI) 0.973, 0.996]), shorter ulcer duration (OR 0.704 [95 % CI 0.574, 0.865]), and dyslipidemia (OR 1.848 [95 % CI 1.052, 3.246]). CONCLUSION: Pro-healing factors associated with VLUs were a smaller ulcer size and a shorter ulcer duration. Dyslipidemia was also associated with improved healing, potentially owing to the use of statins. Patients presenting with poorer-prognosis VLUs should receive more aggressive treatment with earlier referral to vascular surgery.