Development of UroSAM: A Machine Learning Model to Automatically Identify Kidney Stone Composition from Endoscopic Video.

Introduction: Chemical composition analysis is important in prevention counseling for kidney stone disease. Advances in laser technology have made dusting techniques more prevalent, but this offers no consistent way to collect enough material to send for chemical analysis, leading many to forgo this test. We developed a novel machine learning (ML) model to effectively assess stone composition based on intraoperative endoscopic video data. Methods: Two endourologists performed ureteroscopy for kidney stones ≥ 10 mm. Representative videos were recorded intraoperatively. Individual frames were extracted from the videos, and the stone was outlined by human tracing. An ML model, UroSAM, was built and trained to automatically identify kidney stones in the images and predict the majority stone composition as follows: calcium oxalate monohydrate (COM), dihydrate (COD), calcium phosphate (CAP), or uric acid (UA). UroSAM was built on top of the publicly available Segment Anything Model (SAM) and incorporated a U-Net convolutional neural network (CNN). Discussion: A total of 78 ureteroscopy videos were collected; 50 were used for the model after exclusions (32 COM, 8 COD, 8 CAP, 2 UA). The ML model segmented the images with 94.77% precision. Dice coefficient (0.9135) and Intersection over Union (0.8496) confirmed good segmentation performance of the ML model. A video-wise evaluation demonstrated 60% correct classification of stone composition. Subgroup analysis showed correct classification in 84.4% of COM videos. A post hoc adaptive threshold technique was used to mitigate biasing of the model toward COM because of data imbalance; this improved the overall correct classification to 62% while improving the classification of COD, CAP, and UA videos. Conclusions: This study demonstrates the effective development of UroSAM, an ML model that precisely identifies kidney stones from natural endoscopic video data. More high-quality video data will improve the performance of the model in classifying the majority stone composition.

Low-dose fluoroscopy technique drastically decreases patient radiation exposure during percutaneous nephrolithotomy.

Fluoroscopy is essential in percutaneous nephrolithotomy (PCNL) but exposes patients and operating room staff to radiation. We investigated whether a low-dose (LD) protocol could reduce radiation exposure during fluoroscopy-guided access without compromising clinical outcomes. Patients undergoing PCNL with fluoroscopy-guided access at a tertiary care stone center between January 2019 and July 2021 were identified. Prior to September 3, 2020, the Philips Veradius C-arm's default settings were used: standard per-frame dose, 15 pulses per second (PPS) frame rate. After this date, a low-dose protocol was used: reduced per-frame dose, reduced frame rate of 8 PPS for needle puncture and 4 PPS for all other steps. Clinical and radiographical data were retrospectively collected. The primary outcome was cumulative radiation dose. Secondary outcomes were stone-free status (SFS; defined as no fragments ≥ 2 mm) and complications. Multivariate regression analysis was performed. 100 patients were identified; 31 were in the LD group. The LD cohort was exposed to a significantly lower mean cumulative radiation dose of 11.68 mGy compared to 48.88 mGy (p < 0.0001). There were no differences in operative time, fluoroscopy time, stone burden, SFS, or complications. In a multivariable regression model adjusting for several variables, LD protocol was associated with lower radiation dose while skin-to-calyx-distance (STCD) was positively associated with cumulative radiation dose. Low-dose fluoroscopy and decreased frame rate during PCNL decreased radiation exposure fourfold without affecting SFS or complication rates.

Vincristine and bortezomib use distinct upstream mechanisms to activate a common SARM1-dependent axon degeneration program.

Chemotherapy-induced peripheral neuropathy is one of the most prevalent dose-limiting toxicities of anticancer therapy. Development of effective therapies to prevent chemotherapy-induced neuropathies could be enabled by a mechanistic understanding of axonal breakdown following exposure to neuropathy-causing agents. Here, we reveal the molecular mechanisms underlying axon degeneration induced by 2 widely used chemotherapeutic agents with distinct mechanisms of action: vincristine and bortezomib. We showed previously that genetic deletion of SARM1 blocks vincristine-induced neuropathy and demonstrate here that it also prevents axon destruction following administration of bortezomib in vitro and in vivo. Using cultured neurons, we found that vincristine and bortezomib converge on a core axon degeneration program consisting of nicotinamide mononucleotide NMNAT2, SARM1, and loss of NAD+ but engage different upstream mechanisms that closely resemble Wallerian degeneration after vincristine and apoptosis after bortezomib. We could inhibit the final common axon destruction pathway by preserving axonal NAD+ levels or expressing a candidate gene therapeutic that inhibits SARM1 in vitro. We suggest that these approaches may lead to therapies for vincristine- and bortezomib-induced neuropathies and possibly other forms of peripheral neuropathy.

Focused Ultrasound-enabled Brain Tumor Liquid Biopsy.

Although blood-based liquid biopsies have emerged as a promising non-invasive method to detect biomarkers in various cancers, limited progress has been made for brain tumors. One major obstacle is the blood-brain barrier (BBB), which hinders efficient passage of tumor biomarkers into the peripheral circulation. The objective of this study was to determine whether FUS in combination with microbubbles can enhance the release of biomarkers from the brain tumor to the blood circulation. Two glioblastoma tumor models (U87 and GL261), developed by intracranial injection of respective enhanced green fluorescent protein (eGFP)-transduced glioblastoma cells, were treated by FUS in the presence of systemically injected microbubbles. Effect of FUS on plasma eGFP mRNA levels was determined using quantitative polymerase chain reaction. eGFP mRNA were only detectable in the FUS-treated U87 mice and undetectable in the untreated U87 mice (maximum cycle number set to 40). This finding was replicated in GL261 mice across three different acoustic pressures. The circulating levels of eGFP mRNA were 1,500-4,800 fold higher in the FUS-treated GL261 mice than that of the untreated mice for the three acoustic pressures. This study demonstrated the feasibility of FUS-enabled brain tumor liquid biopsies in two different murine glioma models across different acoustic pressures.

Evaluation and selection of anatomic sites for magnetic resonance imaging-guided mild hyperthermia therapy: a healthy volunteer study.

PURPOSE: Since mild hyperthermia therapy (MHT) requires maintaining the temperature within a narrow window (e.g. 40-43 °C) for an extended duration (up to 1 h), accurate and precise temperature measurements are essential for ensuring safe and effective treatment. This study evaluated the precision and accuracy of MR thermometry in healthy volunteers at different anatomical sites for long scan times. METHODS: A proton resonance frequency shift method was used for MR thermometry. Eight volunteers were subjected to a 5-min scanning protocol, targeting chest wall, bladder wall, and leg muscles. Six volunteers were subjected to a 30-min scanning protocol and three volunteers were subjected to a 60-min scanning protocol, both targeting the leg muscles. The precision and accuracy of the MR thermometry were quantified. Both the mean precision and accuracy <1 °C were used as criteria for acceptable thermometry. RESULTS: Drift-corrected MR thermometry measurements based on 5-min scans of the chest wall, bladder wall, and leg muscles had accuracies of 1.41 ± 0.65, 1.86 ± 1.20, and 0.34 ± 0.44 °C, and precisions of 2.30 ± 1.21, 1.64 ± 0.56, and 0.48 ± 0.05 °C, respectively. Measurements based on 30-min scans of the leg muscles had accuracy and precision of 0.56 ± 0.05 °C and 0.42 ± 0.50 °C, respectively, while the 60-min scans had accuracy and precision of 0.49 ± 0.03 °C and 0.56 ± 0.05 °C, respectively. CONCLUSIONS: Respiration, cardiac, and digestive-related motion pose challenges to MR thermometry of the chest wall and bladder wall. The leg muscles had satisfactory temperature accuracy and precision per the chosen criteria. These results indicate that extremity locations may be preferable targets for MR-guided MHT using the existing MR thermometry technique.