Enhancing deep learning pre-trained networks on diabetic retinopathy fundus photographs with SLIC-G.

Diabetic retinopathy disease contains lesions (e.g., exudates, hemorrhages, and microaneurysms) that are minute to the naked eye. Determining the lesions at pixel level poses a challenge as each pixel does not reflect any semantic entities. Furthermore, the computational cost of inspecting each pixel is expensive because the number of pixels is high even at low resolution. In this work, we propose a hybrid image processing method. Simple Linear Iterative Clustering with Gaussian Filter (SLIC-G) for the purpose of overcoming pixel constraints. The SLIC-G image processing method is divided into two stages: (1) simple linear iterative clustering superpixel segmentation and (2) Gaussian smoothing operation. In such a way, a large number of new transformed datasets are generated and then used for model training. Finally, two performance evaluation metrics that are suitable for imbalanced diabetic retinopathy datasets were used to validate the effectiveness of the proposed SLIC-G. The results indicate that, in comparison to prior published works' results, the proposed SLIC-G shows better performance on image classification of class imbalanced diabetic retinopathy datasets. This research reveals the importance of image processing and how it influences the performance of deep learning networks. The proposed SLIC-G enhances pre-trained network performance by eliminating the local redundancy of an image, which preserves local structures, but avoids over-segmented, noisy clips. It closes the research gap by introducing the use of superpixel segmentation and Gaussian smoothing operation as image processing methods in diabetic retinopathy-related tasks.

The Outcomes of App-Based Health Coaching to Improve Dietary Behavior Among Nurses in a Tertiary Hospital: Pilot Intervention Study.

BACKGROUND: At the workplace, health care workers face multiple challenges in maintaining healthy dietary behaviors, which is the major factor behind obesity. A hospital-wide mass health screening exercise showed an increasing trend in the prevalence of obesity and median BMI from 2004 to 2019, as well as a higher crude obesity rate among shift workers. OBJECTIVE: We aimed to evaluate the effectiveness of mobile app-based health coaching and incentives for achieving weight loss from better dietary choices among hospital nurses. METHODS: We conducted a pilot study from June 2019 to March 2020, involving the use of a health-coaching app by 145 hospital nurses over 6 months. Weight and BMI were self-reported, and food scores were calculated. Data among overweight nurses, shift work nurses, and incentive groups were analyzed. RESULTS: A total of 61 nurses were included in the final analysis. Of these 61 nurses, 38 (62%) lost weight. The median percentage weight loss was 1.2% (IQR 0%-2.9%; P<.001), and the median decrease in BMI was 0.35 (IQR -0.15 to 0.82; P<.001), but they were not clinically significant. The median improvement in the food score was 0.4 (IQR 0-0.8). There was no difference between the incentive and nonincentive groups. A total of 49 (34%) participants engaged for ≥8 weeks. CONCLUSIONS: The study demonstrated an association between the use of app-based health coaching and the attainment of some weight loss in nurses, without a significant improvement in the food score. Incentives may nudge on-boarding, but do not sustain engagement.

The adoption of deep learning interpretability techniques on diabetic retinopathy analysis: a review.

Diabetic retinopathy (DR) is a chronic eye condition that is rapidly growing due to the prevalence of diabetes. There are challenges such as the dearth of ophthalmologists, healthcare resources, and facilities that are unable to provide patients with appropriate eye screening services. As a result, deep learning (DL) has the potential to play a critical role as a powerful automated diagnostic tool in the field of ophthalmology, particularly in the early detection of DR when compared to traditional detection techniques. The DL models are known as black boxes, despite the fact that they are widely adopted. They make no attempt to explain how the model learns representations or why it makes a particular prediction. Due to the black box design architecture, DL methods make it difficult for intended end-users like ophthalmologists to grasp how the models function, preventing model acceptance for clinical usage. Recently, several studies on the interpretability of DL methods used in DR-related tasks such as DR classification and segmentation have been published. The goal of this paper is to provide a detailed overview of interpretability strategies used in DR-related tasks. This paper also includes the authors' insights and future directions in the field of DR to help the research community overcome research problems.

Outcomes of an accelerated nonsurgical management protocol for hip fractures in the elderly.

The majority of hip fractures in elderly patients are managed surgically with superior outcomes. However, for patients who refuse surgery or are deemed medically unfit, traction used to be the mainstay of nonsurgical treatment, which is associated with prolonged hospitalization and inpatient complications from immobility. This study, therefore, aims to evaluate the outcomes of an early wheelchair mobilization protocol as an alternative nonsurgical treatment option. This is a retrospective study of 87 elderly patients who were managed nonsurgically for their hip fractures over a 1-year period. The accelerated rehabilitation protocol did not have them on traction but were instead mobilized with assistance as soon as possible after admission. Variables collected electronically include patient demographics, fracture characteristics, inpatient mobilization milestones, inpatient complications, Modified Functional Ambulation Classification (MFAC), Modified Barthel Index (MBI) scores, and radiological findings. Patients who were younger, could sit up earlier and had a shorter length of stay, were able to ambulate better at 6 months ( p value < 0.05). Patients with femoral neck fractures and shorter length of stay had better MFAC scores. A total of 58% of patients with radiological follow-up had displacement of their fractures with age, type of fracture, and length of stay as predictors ( p value < 0.05) The Charlson's score, day to sitting up, and day to transfer affect fracture healing ( p value < 0.05). The mean length of stay was 17 days and the 1-year mortality was 18%. Surgical therapy remains the preferred choice of management for patients with hip fractures. Early wheelchair mobilization leads to a shorter length of stay compared to traditional traction methods and comparable 1-year mortality rates with operative management. Despite this, complication rates remain high and only 48% of patients achieved ambulation by 1 year, with healing in only 24% of fractures.

Clinical Relevance of the Nonvisualized Appendix on Ultrasonography of the Abdomen in Children.

OBJECTIVES: To evaluate the clinical relevance of the nonvisualized appendix on ultrasound imaging in children with right lower quadrant pain. STUDY DESIGN: We reviewed 1359 children admitted for abdominal pain between January and December 2013 who had abdominal ultrasound imaging for right lower quadrant pain. Patients who had scans for genitourinary symptoms or intussusception were excluded from the study. When the appendix was not visualized, secondary signs indicating right lower quadrant inflammatory pathology were noted. RESULTS: Of all admissions for abdominal pain, 810 had ultrasound scans. Thirty-eight did not evaluate the appendix and 131 were excluded for suspected intussusception, leaving 641 reports for children with a median age of 10.8 years (range, 1.3-21.3); 297 were boys (46.3%). There were 17 of 160 patients with a nonvisualized appendix (10.6%) who underwent appendectomy. Of these, 14 had secondary signs on ultrasound imaging and 3 (1.9%) had normal ultrasound reports. The 3 patients with normal ultrasound imaging had computed tomography imaging confirming appendicitis. There were 51 patients with a partially visualized appendix. The segment of appendix that could be seen was normal in 34 patients, none of whom had appendectomy. The remaining 17 had appendectomy, in whom the appendix seemed to be inflamed in 13 and equivocal in 4, all with histologically confirmed appendicitis. Overall, 232 children underwent appendectomy; 58 had no ultrasound imaging done, and 5 had a histologically normal appendix (overall negative appendectomy rate, 2.2%). Only 35 of 1359 patients (0.03%) had computed tomography scans. CONCLUSION: In patients with a nonvisualized appendix on ultrasound imaging and no evidence of secondary inflammatory changes, the likelihood of appendicitis is less than 2%. Generous use of ultrasonography as an adjunct to clinical examination can achieve low negative appendectomy rates without underdiagnosis of acute appendicitis.