Objective: This study aimed to develop and validate a deep learning (DL) algorithm for the quantitative measurement of thoracolumbar (TL) fracture features, and to evaluate its efficacy across varying levels of clinical expertise. Methods: Using the pretrained Mask Region-Based Convolutional Neural Networks model, originally developed for vertebral body segmentation and fracture detection, we fine-tuned the model and added a new module for measuring fracture metrics—compression rate (CR), Cobb angle (CA), Gardner angle (GA), and sagittal index (SI)—from lumbar spine lateral radiographs. These metrics were derived from six-point labeling by 3 radiologists, forming the ground truth (GT). Training utilized 1,000 nonfractured and 318 fractured radiographs, while validations employed 213 internal and 200 external fractured radiographs. The accuracy of the DL algorithm in quantifying fracture features was evaluated against GT using the intraclass correlation coefficient. Additionally, 4 readers with varying expertise levels, including trainees and an attending spine surgeon, performed measurements with and without DL assistance, and their results were compared to GT and the DL model. Results: The DL algorithm demonstrated good to excellent agreement with GT for CR, CA, GA, and SI in both internal (0.860, 0.944, 0.932, and 0.779, respectively) and external (0.836, 0.940, 0.916, and 0.815, respectively) validations. DL-assisted measurements significantly improved most measurement values, particularly for trainees. Conclusion The DL algorithm was validated as an accurate tool for quantifying TL fracture features using radiographs. DL-assisted measurement is expected to expedite the diagnostic process and enhance reliability, particularly benefiting less experienced clinicians.

Objective: This study aimed to develop and validate a deep learning (DL) algorithm for the quantitative measurement of thoracolumbar (TL) fracture features, and to evaluate its efficacy across varying levels of clinical expertise. Methods: Using the pretrained Mask Region-Based Convolutional Neural Networks model, originally developed for vertebral body segmentation and fracture detection, we fine-tuned the model and added a new module for measuring fracture metrics—compression rate (CR), Cobb angle (CA), Gardner angle (GA), and sagittal index (SI)—from lumbar spine lateral radiographs. These metrics were derived from six-point labeling by 3 radiologists, forming the ground truth (GT). Training utilized 1,000 nonfractured and 318 fractured radiographs, while validations employed 213 internal and 200 external fractured radiographs. The accuracy of the DL algorithm in quantifying fracture features was evaluated against GT using the intraclass correlation coefficient. Additionally, 4 readers with varying expertise levels, including trainees and an attending spine surgeon, performed measurements with and without DL assistance, and their results were compared to GT and the DL model. Results: The DL algorithm demonstrated good to excellent agreement with GT for CR, CA, GA, and SI in both internal (0.860, 0.944, 0.932, and 0.779, respectively) and external (0.836, 0.940, 0.916, and 0.815, respectively) validations. DL-assisted measurements significantly improved most measurement values, particularly for trainees. Conclusion The DL algorithm was validated as an accurate tool for quantifying TL fracture features using radiographs. DL-assisted measurement is expected to expedite the diagnostic process and enhance reliability, particularly benefiting less experienced clinicians.

Objective: This study aimed to develop and validate a deep learning (DL) algorithm for the quantitative measurement of thoracolumbar (TL) fracture features, and to evaluate its efficacy across varying levels of clinical expertise. Methods: Using the pretrained Mask Region-Based Convolutional Neural Networks model, originally developed for vertebral body segmentation and fracture detection, we fine-tuned the model and added a new module for measuring fracture metrics—compression rate (CR), Cobb angle (CA), Gardner angle (GA), and sagittal index (SI)—from lumbar spine lateral radiographs. These metrics were derived from six-point labeling by 3 radiologists, forming the ground truth (GT). Training utilized 1,000 nonfractured and 318 fractured radiographs, while validations employed 213 internal and 200 external fractured radiographs. The accuracy of the DL algorithm in quantifying fracture features was evaluated against GT using the intraclass correlation coefficient. Additionally, 4 readers with varying expertise levels, including trainees and an attending spine surgeon, performed measurements with and without DL assistance, and their results were compared to GT and the DL model. Results: The DL algorithm demonstrated good to excellent agreement with GT for CR, CA, GA, and SI in both internal (0.860, 0.944, 0.932, and 0.779, respectively) and external (0.836, 0.940, 0.916, and 0.815, respectively) validations. DL-assisted measurements significantly improved most measurement values, particularly for trainees. Conclusion The DL algorithm was validated as an accurate tool for quantifying TL fracture features using radiographs. DL-assisted measurement is expected to expedite the diagnostic process and enhance reliability, particularly benefiting less experienced clinicians.

Objective: This study aimed to develop and validate a deep learning (DL) algorithm for the quantitative measurement of thoracolumbar (TL) fracture features, and to evaluate its efficacy across varying levels of clinical expertise. Methods: Using the pretrained Mask Region-Based Convolutional Neural Networks model, originally developed for vertebral body segmentation and fracture detection, we fine-tuned the model and added a new module for measuring fracture metrics—compression rate (CR), Cobb angle (CA), Gardner angle (GA), and sagittal index (SI)—from lumbar spine lateral radiographs. These metrics were derived from six-point labeling by 3 radiologists, forming the ground truth (GT). Training utilized 1,000 nonfractured and 318 fractured radiographs, while validations employed 213 internal and 200 external fractured radiographs. The accuracy of the DL algorithm in quantifying fracture features was evaluated against GT using the intraclass correlation coefficient. Additionally, 4 readers with varying expertise levels, including trainees and an attending spine surgeon, performed measurements with and without DL assistance, and their results were compared to GT and the DL model. Results: The DL algorithm demonstrated good to excellent agreement with GT for CR, CA, GA, and SI in both internal (0.860, 0.944, 0.932, and 0.779, respectively) and external (0.836, 0.940, 0.916, and 0.815, respectively) validations. DL-assisted measurements significantly improved most measurement values, particularly for trainees. Conclusion The DL algorithm was validated as an accurate tool for quantifying TL fracture features using radiographs. DL-assisted measurement is expected to expedite the diagnostic process and enhance reliability, particularly benefiting less experienced clinicians.

Objective: This study aimed to develop and validate a deep learning (DL) algorithm for the quantitative measurement of thoracolumbar (TL) fracture features, and to evaluate its efficacy across varying levels of clinical expertise. Methods: Using the pretrained Mask Region-Based Convolutional Neural Networks model, originally developed for vertebral body segmentation and fracture detection, we fine-tuned the model and added a new module for measuring fracture metrics—compression rate (CR), Cobb angle (CA), Gardner angle (GA), and sagittal index (SI)—from lumbar spine lateral radiographs. These metrics were derived from six-point labeling by 3 radiologists, forming the ground truth (GT). Training utilized 1,000 nonfractured and 318 fractured radiographs, while validations employed 213 internal and 200 external fractured radiographs. The accuracy of the DL algorithm in quantifying fracture features was evaluated against GT using the intraclass correlation coefficient. Additionally, 4 readers with varying expertise levels, including trainees and an attending spine surgeon, performed measurements with and without DL assistance, and their results were compared to GT and the DL model. Results: The DL algorithm demonstrated good to excellent agreement with GT for CR, CA, GA, and SI in both internal (0.860, 0.944, 0.932, and 0.779, respectively) and external (0.836, 0.940, 0.916, and 0.815, respectively) validations. DL-assisted measurements significantly improved most measurement values, particularly for trainees. Conclusion The DL algorithm was validated as an accurate tool for quantifying TL fracture features using radiographs. DL-assisted measurement is expected to expedite the diagnostic process and enhance reliability, particularly benefiting less experienced clinicians.

Purpose: Endoscopic treatment and laparoscopic surgery are minimally invasive options for early treatment of colorectal cancer, however, more evidence of the long-term outcomes between the two procedures is needed to guide clinical decisions. Therefore, this study aimed to compare the oncologic outcomes between endoscopic and laparoscopic treatment for early colorectal cancer. Methods: The study group included 60 patients who underwent endoscopic treatment and 38 patients who underwent laparoscopic surgery for early colorectal adenocarcinoma between January 2010 and December 2013 at a single study site. Results: Histopathological diagnoses showed that 43 (78.3%) carcinomas in the endoscopic submucosal dissection group were mucosal to sm1, 13 (21.7%) were sm2 or deeper, and 17 high-risk cases (28.3%) in the endoscopic group underwent additional surgery. The median operation time, time to sips of water, and length of hospital stay were significantly shorter in the endoscopic group than in the laparoscopic group. The overall survival rates of patients in the endoscopic group and laparoscopic groups were 91.5% and 87.4%, respectively (P=0.391), and the disease-free survival rates were 90.4% and 87.4% (P=0.614), respectively. Systemic recurrences occurred in two patients (1.6%) in the endoscopic group and one patient (2.0%) in the laparoscopic group. Local recurrence combined with systemic recurrence in one patient (0.8%) in the endoscopic group. Conclusion Endoscopic resection for early colorectal cancer can be performed safely with better short-term outcomes and comparable longterm oncological outcomes compared to laparoscopic surgery.

Background/Aims: Renal ischemia followed by reperfusion (I/R) is a leading cause of acute kidney injury (AKI), which is closely associated with high morbidity and mortality. Studies have shown that induced pluripotent stem cell (iPSC)-derived mesenchymal stem cells (iMSCs) exert powerful therapeutic effects in renal ischemia. However, the efficacy of iMSC-derived exosomes (iExo) on I/R injuries remains largely unknown. Methods: Human iPSCs were differentiated into iMSCs using a modified one-step method. Ultrafiltration, combined with purification, was used to isolate iExo from iMSCs. iExo was administered following I/R injury in a mouse model. The effect of iExo on I/R injury was assessed through changes in renal function, histology, and expression of oxidative stress, inflammation, and apoptosis markers. Further, we evaluated its association with the extracellular signal-regulated kinase (ERK) 1/2 signaling pathway. Results: Mice subjected to I/R injury exhibited typical AKI patterns; serum creatinine level, tubular necrosis, apoptosis, inflammatory cytokine production, and oxidative stress were markedly increased compared to sham mice. However, treatment with iExo attenuated these changes, significantly improving renal function and tissue damage, similar to the renoprotective effects of iMSCs on I/R injury. Significant induction of activated ERK 1/2 signaling molecules was observed in mice treated with iExo compared to those in the I/R injury group. Conclusions The present study demonstrates that iExo administration ameliorated renal damage following I/R, suggesting that iMSC-derived exosomes may provide a novel therapeutic approach for AKI treatment.

Purpose: Alpha-fetoprotein (AFP) is a prognostic marker for hepatocellular carcinoma (HCC). We investigated the prognostic value of AFP levels in patients who achieved complete response (CR) to transarterial chemoembolization (TACE) for HCC. Materials and Methods: Between 2005 and 2018, 890 patients with HCC who achieved a CR to TACE were recruited. An AFP responder was defined as a patient who showed elevated levels of AFP (>10 ng/mL) during TACE, but showed normalization or a >50% reduction in AFP levels after achieving a CR. Results: Among the recruited patients, 569 (63.9%) with naïve HCC and 321 (36.1%) with recurrent HCC after complete resection were treated. Before TACE, 305 (34.3%) patients had multiple tumors, 219 (24.6%) had a maximal tumor size >3 cm, and 22 (2.5%) had portal vein tumor thrombosis. The median AFP level after achieving a CR was 6.36 ng/mL. After a CR, 473 (53.1%) patients experienced recurrence, and 417 (46.9%) died [median progression-free survival (PFS) and overall survival (OS) of 16.3 and 62.8 months, respectively]. High AFP levels at CR (>20 ng/mL) were independently associated with a shorter PFS [hazard ratio (HR)=1.403] and OS (HR=1.284), together with tumor multiplicity at TACE (HR=1.518 and 1.666, respectively). AFP non-responders at CR (76.2%, n=359 of 471) showed a shorter PFS (median 10.5 months vs. 15.5 months, HR=1.375) and OS (median 41.4 months vs. 61.8 months, HR=1.424) than AFP responders (all p=0.001). Conclusion High AFP levels and AFP non-responders were independently associated with poor outcomes after TACE. AFP holds clinical implications for detailed risk stratification upon achieving a CR after TACE.