Automatic Classification of Nodules from 2D Ultrasound Images Using Deep Learning Networks.

OBJECTIVE: In clinical practice, thyroid nodules are typically visually evaluated by expert physicians using 2D ultrasound images. Based on their assessment, a fine needle aspiration (FNA) may be recommended. However, visually classifying thyroid nodules from ultrasound images may lead to unnecessary fine needle aspirations for patients. The aim of this study is to develop an automatic thyroid ultrasound image classification system to prevent unnecessary FNAs. METHODS: An automatic computer-aided artificial intelligence system is proposed for classifying thyroid nodules using a fine-tuned deep learning model based on the DenseNet architecture, which incorporates an attention module. The dataset comprises 591 thyroid nodule images categorized based on the Bethesda score. Thyroid nodules are classified as either requiring FNA or not. The challenges encountered in this task include managing variability in image quality, addressing the presence of artifacts in ultrasound image datasets, tackling class imbalance, and ensuring model interpretability. We employed techniques such as data augmentation, class weighting, and gradient-weighted class activation maps (Grad-CAM) to enhance model performance and provide insights into decision making. RESULTS: Our approach achieved excellent results with an average accuracy of 0.94, F1-score of 0.93, and sensitivity of 0.96. The use of Grad-CAM gives insights on the decision making and then reinforce the reliability of the binary classification for the end-user perspective. CONCLUSIONS: We propose a deep learning architecture that effectively classifies thyroid nodules as requiring FNA or not from ultrasound images. Despite challenges related to image variability, class imbalance, and interpretability, our method demonstrated a high classification accuracy with minimal false negatives, showing its potential to reduce unnecessary FNAs in clinical settings.

Impact of a Tripartite Collaboration between Oncologist, Pharmacist and Diabetologist in the Management of Patients with Diabetes Starting Chemotherapy: The ONCODIAB Trial.

BACKGROUND: Diabetes negatively impacts cancer prognosis. The objective of this work was to evaluate a tripartite oncologist-pharmacist-diabetologist collaboration in the management of patients with diabetes starting chemotherapy. PATIENTS AND METHODS: The prospective ONCODIAB study (NCT04315857) included 102 adults with diabetes starting chemotherapy by whom a continuous glucose monitoring device was worn for fourteen days from the first day of the first and second chemotherapy cycles. The primary outcome was to assess pharmacist and diabetologist interventions. The secondary outcome was to evaluate the impact of the ONCODIAB follow-up on individualized patient glycemic targets at 6 months. RESULTS: A total of 191 (2 per patient) were made either by clinical pharmacists (n = 95) or diabetologists (n = 96) during the first two chemotherapy cycles. The anatomic therapeutic chemical drug classes most frequently involved in pharmacist interventions were cardiovascular system (23%), alimentary tract and metabolism (22%), and anti-infectives for systemic use (14%). Diabetologists modified the antidiabetic treatment in 58 (62%) of patients: dose reduction (34%), drug discontinuation (28%), drug addition (24%), and dose increase (15%). Glycated hemoglobin decreased from 7.6 ± 1.7% at baseline to 7.1 ± 1.1% at 6 months (p = 0.02). Compared to individualized targets, HbA1c was higher, in the interval, or lower in 29%, 44%, and 27% of patients at baseline vs. in 8%, 70%, and 22% of patients at 6 months, respectively (p < 10-3). CONCLUSIONS: In our study, a close collaboration between oncologists, pharmacists, and diabetologists helped by continuous glucose monitoring led to overall medication optimization and better glycemic control in patients with diabetes starting chemotherapy.

Glycemic control in people with diabetes treated with cancer chemotherapy: contribution of continuous glucose monitoring.

AIMS: The aim of our study was to assess, with Continuous Glucose Monitoring (CGM), exhaustive information on the glucose profile in people with diabetes starting chemotherapy. We also evaluated the adaptation of glucose-lowering drugs following analysis of CGM recordings. METHODS: Eighty-five people with diabetes starting chemotherapy were included in the ONCODIAB study. A CGM was worn for up to fourteen days in blinded mode before and after the diabetologist's intervention to evaluate the impact of modifying the glucose-lowering drugs. RESULTS: Time spent in range was 67.2 ± 24.2%. Time below the target glucose range (TBR) (< 70 mg/dl) was 8.9% in all the study population. TBR was significantly higher in patients treated with at least one drug due to the risk of hypoglycemia compared to the others (11.5% vs. 4.4%, p = 0.009). Sixty-five patients had available sensor data for the two recordings. Forty-one patients (51.9%) saw a decrease in their antidiabetic treatment after the diabetologist's intervention guided by the first CGM recording. We observed a significant reduction in the time spent below the target glucose range (70-55 mg/dl) between the two CGM recordings (10.3 ± 14.6% vs. 6.3 ± 9.4%, p = 0.016 and 3.8 ± 8.4% vs. 1.2 ± 2.9%, p = 0.012, respectively). CONCLUSIONS: CGM use in blinded mode could be an interesting tool to reduce the risk of hypoglycemia in people with diabetes starting chemotherapy. Our findings fully support the recommendation that assessing hypoglycemia risk should be mandatory in patients with diabetes before starting chemotherapy.

Fewer Type A personality traits in type 2 diabetes patients with diabetic foot ulcer.

AIM: Type A personality-characterized by time urgency, strong drive, and a need for achievement and competitiveness-has been shown to be associated with reduced mortality in patients with diabetes. However, it is not known whether a Type A personality might protect against diabetic foot ulcer (DFU). This prompted our present analysis of the association between Type A personality and DFU. METHODS: The Bortner Scale questionnaire was used to assess Type A personality in 386 patients with type 2 diabetes (T2D), including 104 patients also presenting with, and 282 presenting without, DFU. Additional questionnaires were used to assess perceived stress and depression. RESULTS: Type A Bortner scores were significantly lower in T2D patients with vs without DFU (166.64 ± 38.76 vs 178.79 ± 36.61, respectively; P = 0.005). In patients with DFU, the prevalence of Type A personality traits was significantly lower than in those without DFU (48% vs 64.5%, respectively; P = 0.005) whereas, in contrast, Type B personality traits (the opposite of Type A) were more prevalent (52% vs 35.5%, respectively; P = 0.005). On multivariate analysis, Type A Bortner scores were negatively associated with DFU (P = 0.008) independently of age, gender, BMI, depression scores or perceived stress. CONCLUSION: The Type A personality, characterized by competitiveness and a need for achievement, is significantly less frequently seen in T2D patients with DFU. On the other hand, the Type B personality is much more prevalent in such patients. It may be that the Type B personality, which is characterized by fewer problem-focused coping strategies and a decreased adherence to care, might favour the development of DFU.

Successful Control of Hypoglycemia with Pasireotide LAR in a Patient with Inappropriate Insulin Secretion.

INTRODUCTION: Inappropriate insulin secretion could be due to several diseases. Nesidioblastosis is characterized by diffuse hyperplasia of pancreatic beta cells, causing organic hypoglycemia. No pancreatic lesions are found on the imaging of patients with this condition. Diazoxide is used as a first-line treatment but can be poorly tolerated because of its side effects, and therapeutic failure is possible. Somatostatin analogues have limited efficacy because of their poor affinity to somatostatin (SST) receptors. Pasireotide is a somatostatin analogue with a much higher affinity to SST receptors, especially SST5, and it could thus be more efficient for treating nesidioblastosis-related hypoglycemia. OBSERVATION: A 56 years-old diabetic woman had symptoms of hypoglycemia, persistent after treatment's withdrawal. A fasting test authentify an organic hypoglycemia, at 34mg/dL, a plasma insulin level at 6mUI/L above the 5 mU/L threshold, a C-peptide level at 1.9 ng/mL above the threshold of 0.6, and an insulin/C-peptide ratio 0.066, below the threshold of 1. No lesions were found on CT-scan or endoscopic ultrasound. Somatostatin receptor scintigraphy was also negative. Diazoxide and octreotide failed to improve the recurrence of hypoglycemia episodes. With pasireotide LAR, hypoglycemia disappeared and glycemia increased. Hyperglycemia was controlled with sitagliptin. The patient has now been treated with pasireotide LAR for two years, with no more episode of hypoglycemia until now. DISCUSSION: We present the first case of nesidioblastosis treatment with pasireotide LAR, with success. Patients diagnosed with nesidioblastosis and diazoxide-resistant hypoglycemia, or who experience difficulties with other treatments, could use pasireotide LAR in conjunction with glycemia monitoring, particularly if they are diabetic.

Increased body fat mass reduces the association between fructosamine and glycated hemoglobin in obese type 2 diabetes patients.

Obesity is increasing in patients with type 2 diabetes. A possible reduced association between fructosamine and glycated hemoglobin (HbA1c) in obese individuals has been previously discussed, but this has never been specifically evaluated in type 2 diabetes, and the potential influence of body fat mass and fat distribution has never been studied. We studied 112 type 2 diabetes patients with assessment of fat mass, liver fat and fat distribution. Patients with body mass index (BMI) above the median (34.9 kg/m2 ), versus BMI below the median, had a correlation coefficient between fructosamine and HbA1c significantly reduced (r = 0.358 vs r = 0.765). In the whole population, fructosamine was correlated negatively with BMI and fat mass. In multivariate analysis, fructosamine was associated with HbA1c (positively) and fat mass (negatively), but not with BMI, liver fat or fat distribution. The association between fructosamine and HbA1c is significantly reduced in the most obese type 2 diabetes patients, and this is mostly driven by increased fat mass.