Deep learning on pre-procedural computed tomography and clinical data predicts outcome following stroke thrombectomy.

BACKGROUND: Deep learning using clinical and imaging data may improve pre-treatment prognostication in ischemic stroke patients undergoing endovascular thrombectomy (EVT). METHODS: Deep learning models were trained and tested on baseline clinical and imaging (CT head and CT angiography) data to predict 3-month functional outcomes in stroke patients who underwent EVT. Classical machine learning models (logistic regression and random forest classifiers) were constructed to compare their performance with the deep learning models. An external validation dataset was used to validate the models. The MR PREDICTS prognostic tool was tested on the external validation set, and its performance was compared with the deep learning and classical machine learning models. RESULTS: A total of 975 patients (550 men; mean±SD age 67.5±15.1 years) were studied with 778 patients in the model development cohort and 197 in the external validation cohort. The deep learning model trained on baseline CT and clinical data, and the logistic regression model (clinical data alone) demonstrated the strongest discriminative abilities for 3-month functional outcome and were comparable (AUC 0.811 vs 0.817, Q=0.82). Both models exhibited superior prognostic performance than the other deep learning (CT head alone, CT head, and CT angiography) and MR PREDICTS models (all Q<0.05). CONCLUSIONS: The discriminative performance of deep learning for predicting functional independence was comparable to logistic regression. Future studies should focus on whether incorporating procedural and post-procedural data significantly improves model performance.

Transarterial Radioembolization for Hepatic Metastases of Pancreatic Adenocarcinoma: A Systematic Review.

PURPOSE: To assess the safety and effectiveness of transarterial radioembolization (TARE) in the treatment of hepatic metastases from pancreatic ductal adenocarcinoma (PDAC). MATERIALS AND METHODS: A systematic search of the Embase and MEDLINE databases was conducted using keywords and Medical Subject Headings terms related to TARE and hepatic metastases from PDAC. Observational studies and clinical trials reporting overall survival (OS), hepatic progression-free survival (hPFS), or tumor response after TARE were included. RESULTS: Eight studies, comprising 145 patients with metastatic PDAC, met the inclusion criteria. No randomized controlled trials were identified, and 4 studies were prospective. Forty-four (30.3%) patients underwent previous pancreatic resection, and 66 (45.5%) had extrahepatic metastases at the time of TARE. Most studies (n = 6) used resin microspheres for TARE. The pooled disease control rate was 69.4% at a median of 3 months. The median OS from the time of TARE ranged from 3.7 to 9 months. The median hPFS ranged from 2.4 to 5.2 months. There were 31 Grade 3-4 biochemical toxicities and 4 treatment-related deaths. CONCLUSIONS: The role of TARE in patients with hepatic metastases from PDAC remains unclear owing to low patient numbers, limited prospective data, and heterogeneity in the study design. Further prospective studies are required to evaluate the role of TARE in carefully selected patients with liver-only metastatic disease.

Accuracy of ultrasound in the estimation of customised birth weight in a public hospital service.

INTRODUCTION: Studies have shown that ultrasound estimated foetal weight (EFW) in small for gestational age (SGA) babies tends to be less-accurate when compared to appropriate (AGA) and large (LGA) for gestational age babies. We aimed to analyse the accuracy of ultrasound EFW overall, and by customised birth weight centile category (severe SGA, SGA, AGA, LGA). Also, the accuracy of estimating the centile category using calculated customised EFW centiles. METHODS: We performed a retrospective study of pregnant women between 20-43 weeks gestation who underwent ultrasound within 7 days of delivery at a large tertiary maternity unit between January 2018 and December 2020. Stillbirths, major foetal anomalies and multiple pregnancies were excluded. The EFW and birth weight were compared, and an accurate estimate defined as ≤15% difference. The customised EFW and birth weight centiles were calculated and used to analyse the accuracy of category prediction. RESULTS: Of 2061 foetuses included, 92% (n = 1902) were born weighing within 15% of their EFW. Accuracy was not affected by maternal BMI, ethnicity, parity or gestation. 87% of SGA babies were within 15% of their EFW. Ultrasound sensitivity for SGA was 51% (95% CI: 46-55%). The specificity and positive predictive values were 97% (95% CI: 96-98%) and 87% (95% CI: 82-90%) respectively. CONCLUSION: The accuracy of Ultrasound EFW overall is good, however, is reduced in SGA babies whose EFW and birth weight centile categories tended to be overestimated. The high specificity for SGA supports monitoring with a lowered threshold to intervene in pregnancies identified by ultrasound as SGA.

Automated pneumothorax triaging in chest X-rays in the New Zealand population using deep-learning algorithms.

INTRODUCTION: The primary aim was to develop convolutional neural network (CNN)-based artificial intelligence (AI) models for pneumothorax classification and segmentation for automated chest X-ray (CXR) triaging. A secondary aim was to perform interpretability analysis on the best-performing candidate model to determine whether the model's predictions were susceptible to bias or confounding. METHOD: A CANDID-PTX dataset, that included 19,237 anonymized and manually labelled CXRs, was used for training and testing candidate models for pneumothorax classification and segmentation. Evaluation metrics for classification performance included Area under the receiver operating characteristic curve (AUC-ROC), sensitivity and specificity, whilst segmentation performance was measured using mean Dice and true-positive (TP)-Dice coefficients. Interpretability analysis was performed using Grad-CAM heatmaps. Finally, the best-performing model was implemented for a triage simulation. RESULTS: The best-performing model demonstrated a sensitivity of 0.93, specificity of 0.95 and AUC-ROC of 0.94 in identifying the presence of pneumothorax. A TP-Dice coefficient of 0.69 is given for segmentation performance. In triage simulation, mean reporting delay for pneumothorax-containing CXRs is reduced from 9.8 ± 2 days to 1.0 ± 0.5 days (P-value < 0.001 at 5% significance level), with sensitivity 0.95 and specificity of 0.95 given for the classification performance. Finally, interpretability analysis demonstrated models employed logic understandable to radiologists, with negligible bias or confounding in predictions. CONCLUSION: AI models can automate pneumothorax detection with clinically acceptable accuracy, and potentially reduce reporting delays for urgent findings when implemented as triaging tools.

The impact of antibiotic pre-treatment on diagnostic yield of CT-guided biopsy for spondylodiscitis: A multi-centre retrospective study and meta-analysis.

INTRODUCTION: Patients with suspected spondylodiscitis often undergo CT-guided biopsy to identify a causative microbiological organism. Antibiotic pre-treatment has been postulated as a cause for a negative biopsy, although previous clinical studies have been heterogenous with a meta-analysis suggesting no effect. The aim of this study was to assess the impact of antibiotic pre-treatment on microbiological yield. METHODS: Retrospective review of consecutive adult patients undergoing CT-guided biopsy for suspected spondylodiscitis in two tertiary centres between 2010 and 2016. Demographic, procedural and clinical data were collected. Antibiotic pre-treatment was ascertained from patient drug charts. RESULTS: Over the 6-year period, 104 biopsies in 104 patients were included. 51% had a positive microbiological yield at CT-guided biopsy, with the most common isolated organism being Staphylococcus aureus (10.6%). Over two thirds of patients (69.3%) were off antibiotics at time of biopsy. There was no significant difference in microbiological yield in those patients on versus off antibiotics (48.2% vs 54.2%, P = 0.55). 10.6% patients had a final diagnosis of Mycobacterium tuberculosis spondylodiscitis, and this organism was significantly associated with a positive microbiological yield (90.9% vs 46.2%, P = 0.01). There was an inverse association between the presence of fever and sepsis with positive microbiological yield. CONCLUSIONS: CT-guided biopsy in suspected spondylodiscitis obtains a positive microbiological yield in about half of patients. This was significantly higher in patients diagnosed with tuberculosis spondylodiscitis, but there was no significant difference with antibiotic pre-treatment. Therefore, antibiotic pre-treatment should not preclude clinicians from pursuing a microbiological sample through CT-guided biopsy.

Benign Posttraumatic Pseudopneumoperitoneum.

OBJECTIVE: The objective of our study was to retrospectively determine the anatomic distribution of chest wall ectopic gas resembling pneumoperitoneum (i.e., pseudopneumoperitoneum) and its relationship with trauma mechanisms and clinical outcomes using CT. MATERIALS AND METHODS: Investigators from two separate trauma referral centers screened 492 chest, abdomen, and pelvis CT examinations of patients who had sustained any form of trauma between 2010 and 2015. After excluding 186 patients with recognized causes of ectopic gas, 306 patients (211 men and 95 women; mean age, 44.5 years; range, 6-95 years) remained for analysis by two radiology residents in center 1 and a radiology resident in center 2. Positive cases were reviewed by all investigators, including an experienced fellowship-trained abdominal radiologist. The anatomic location of the pseudopneumoperitoneum, injury severity score, trauma velocity (high speed vs low or unknown speed), trauma mechanism, clinical findings on follow-up, and exploratory laparotomy data were collected for patients with pseudopneumoperitoneum. Two hundred consecutive nontrauma CT examinations from 2015 were selected as control cases by a resident in center 1. The t test and chi-square test were used for determining associations. RESULTS: Pseudopneumoperitoneum was identified in 5.2% of patients, occurring bilaterally adjacent to the lower six costochondral junctions, and was significantly more common with high-velocity trauma than with low-velocity trauma (p = 0.010). None of the patients with pseudopneumoperitoneum had evidence of perforated hollow viscus at surgery (n = 2) or on clinical follow-up (n = 14). No patients had unnecessary surgery due to pseudopneumoperitoneum. CONCLUSION: Pseudopneumoperitoneum is a posttraumatic phenomenon centered near the lower six costochondral junctions. Recognizing these findings may help prevent unnecessary laparotomy in the trauma setting.

Do high risk patients alter their lifestyle to reduce risk of colorectal cancer?

BACKGROUND: Colorectal cancer (CRC) may be reduced by healthy lifestyle behaviours. We determined the extent of self-reported lifestyle changes in people at increased risk of CRC, and the association of these reports with anxiety, risk and knowledge-based variables. METHODS: We randomly selected 250 participants who had undergone surveillance colonoscopy for family history of CRC. A telephone interview was conducted, recording demographics and family history. Self-reported lifestyle change due to thoughts about CRC across a range of dietary and lifestyle variables was assessed on a four-point scale. Participants' perceptions of the following were recorded: risk factor knowledge, personal risk, and worry due to family history. General anxiety was assessed using the GAD-7 scale. Ordinal logistic regression was used to calculate adjusted results. RESULTS: There were 148 participants (69% response). 79.7% reported at least one healthy change. Change in diet and physical activity were most frequently reported (fiber, 63%; fruit and vegetables, 54%; red meat, 47%; physical activity, 45%), with consumption of tobacco, alcohol, and body weight less likely (tobacco, 25%; alcohol, 26%; weight 31%). People were more likely to report healthy change with lower levels of generalized anxiety, higher worry due to family history, or greater perceived knowledge of CRC risk factors. Risk perception and risk due to family history were not associated with healthy changes. CONCLUSIONS: Self-reported lifestyle changes due to thoughts about CRC were common. Lower general anxiety levels, worries due to family history, and perceived knowledge of risk factors may stimulate healthy changes.

Perceived risks and benefits of surveillance colonoscopy in people undergoing surveillance for family history of colorectal cancer.

AIM: To determine perceived risks and benefits of colonoscopy surveillance among patients undergoing surveillance colonoscopy due to family history of colorectal cancer (CRC). METHOD: We conducted a standardised phone interview of 250 randomly selected people who had undergone surveillance colonoscopy at Dunedin Hospital. We assessed perceptions of colonoscopy effectiveness, complication rates, and CRC risk. RESULTS: We included 148 (69%) participants. Most felt well informed about surveillance (66.7%), but many wanted further information (63.2%). Most accurately estimated complication rates (discomfort: Common/Uncommon 70.3%; pain: Rare/Uncommon 58.8%; bleeding: Rare/Uncommon 72.3%; perforation: Very rare 62.8%), and benefits (mean reduction in risk of CRC 72.6% and death 76.2%). Most (55.1%) over-estimated their CRC risk. 13.5% thought perforation never occurred, and 12.8% thought colonoscopy either completely prevented, or had no effect on risk of developing or dying from CRC. Patients giving unrealistic estimates had similar demographics and clinical variables to the wider cohort. CONCLUSION: Our results suggest current surveillance education adequately communicates risks and benefits to most patients. A minority have unrealistic views and further education may be indicated.

Assessment and management of leg ulcers in the community and an outpatient clinic.

AIM: To compare the assessment and treatment of leg ulcers seen in the community and subsequently reviewed in an outpatient clinic, to the New Zealand Guidelines. METHODS: An observational study including consecutive patients presenting to vascular surgery outpatients with at least one leg ulcer. Outcomes included the clinical descriptions of ulcers, use of an Ankle Brachial Index (ABI) test and compression therapy for mixed and venous ulcers. RESULTS: The study included seventy-six patients. Every ulcer had an adequate clinical description. An ABI investigation was carried out in 9.1% and 66.7% of the patients in the community and outpatient clinic, respectively. Among 31 patients with venous or mixed ulcers in the community, 7 (22.6%) were initiated on compression therapy, and 1 (3.2%) on compression bandaging. Following outpatient clinic appointments, 29 (76.3%) of the 38 patients diagnosed with venous or mixed ulcers were on compression therapy with 20 (52.6%) on compression bandaging. CONCLUSION: There are low rates of ABI measurements and initiation of compression therapy for patients with leg ulcers in the Otago regional community. This may be due to low accessibility to expertise in ABI measurements delaying initiation of compression therapy. Innovations in facilitating ABI investigation in the community and promoting the use of compression therapy are indicated.