Objective: The artificial intelligence competition in healthcare was organized for the first time at the annual aviation, space, and technology festival (TEKNOFEST), Istanbul/Türkiye, in September 2021. In this article, the data set preparation and competition processes were explained in detail; the anonymized and annotated data set is also provided via official website for further research. Materials and Methods: Data set recorded over the period covering 2019 and 2020 were centrally screened from the e-Pulse and Teleradiology System of the Republic of Türkiye, Ministry of Health using various codes and filtering criteria. The data set was anonymized. The data set was prepared, pooled, curated, and annotated by 7 radiologists. The training data set was shared with the teams via a dedicated file transfer protocol server, which could be accessed using private usernames and passwords given to the teams under a nondisclosure agreement signed by the representative of each team. Results: The competition consisted of 2 stages. In the first stage, teams were given 192 digital imaging and communications in medicine images that belong to 1 of 3 possible categories namely, hemorrhage, ischemic, or non-stroke. Teams were asked to classify each image as either stroke present or absent. In the second stage of the competition, qualifying 36 teams were given 97 digital imaging and communications in medicine images that contained hemorrhage, ischemia, or both lesions. Among the employed methods, Unet and DeepLabv3 were the most frequently observed ones. Conclusion: Artificial intelligence competitions in healthcare offer good opportunities to collect data reflecting various cases and problems. Especially, annotated data set by domain experts is more valuable. Cite this article as: Koç U, Akçapınar Sezer E, Alper Özkaya Y, et al. Artificial intelligence in healthcare competition (TEKNOFEST-2021): Stroke data set. Eurasian J Med., 2022;54(3):248-258.

Magnetic resonance imaging (MRI) and ultrasonography (US) are commonly used in the pre-surgery determination of tumor size and the follow-up of breast cancer patients treated with neoadjuvant chemotherapy (NAC). The aim of this study was to compare the efficiency of preoperative MRI and US in tumor size evaluation of patients with breast cancer after NAC to guide clinicians on the appropriate treatment plan. The study included a total of 75 patients who had undergone radiological follow-up, surgical treatment and pathological examination in our hospital between 2013 and 2016. Of these, 28 patients were followed-up with MRI and 47 with US. The dimension evaluations in pathology examination and on both MRI and US were based on the longest dimension of the tumor. There was no statistically significant difference between the tumor size measured pathologically and the size measured preoperatively on MRI (p=0.379). The tumor size measured on US before surgery was significantly smaller than the size measured in pathology (p=0.004). MRI did not overestimate by more than 10 mm in any patient, whereas US overestimated in 4 patients (8.6%). The correlation coefficient of MRI was higher than that of US (0.927 and 0.687, respectively). MRI is superior to US in preoperative tumor size evaluation of patients receiving NAC.

Background/Aims: Pancreatic steatosis (PS) is a subject of current interest and its prevalence has been reported to range from 16.1% to 30.7% using various radiological methods. This study aimed to evaluate PS prevalence with non-contrast computed tomography (CT). Materials and Methods: The non-contrast CT scans taken in 2016 and 2017 in our hospital were retrospectively screened. A total of 637 cases (320 males, 317 females) were included in the study. CT number measurements were performed from three anatomic regions of the pancreas using regions of interest (ROI) of approximately 1 cm2. The cases with a <0.7 ratio of the pancreatic over splenic CT number were accepted as quantitatively steatosis-positive. Anthropometric evaluations were undertaken by determining various parameters defined on CT. Results: PS was determined visually in 30.6% of the males and 29% of the females, and quantitatively in 32.8% and 30.6%, respectively. A positive agreement was determined between the quantitative and visual evaluations of steatosis (Cohen’s kappa coefficient=0.587, p<0.001). Although PS was seen to be mostly diffuse, the tail region of the pancreas was determined to be the area with most steatosis. Conclusion: PS is usually overlooked in radiology practice but it has a clinical presentation with non-insignificant prevalence. Current radiological methods are adequate in the evaluation of PS. The determination of the cut-off values for various criteria on non-contrast CT can provide more objective evaluations.

Pathologies of the posterior labrocapsular structures of the shoulder joint are far less common than anterior labrocapsuloligamentous lesions. Most of these pathologies have been associated with traumatic posterior dislocation. A smaller portion of the lesions include posterior extension of superior labral anteroposterior lesions, posterior superior internal impingement, and damage to the posterior band of the inferior glenohumeral ligament. Labrocapsular anatomic variations of the posterior shoulder joint can mimic labral pathology on conventional MR and occasionally on MR arthrographic images. Knowledge of this variant anatomy is key to interpreting MR images and studying MR arthrography of the posterior labrocapsular structure to avoid misdiagnosis and unnecessary surgical procedures. In this article, we review normal and variant anatomy of the posterior labrocapsular structure of the shoulder joint based on MR arthrography and discuss how to discriminate normal anatomic variants from labrocapsular damage.