Khosravi, Bardia

Background: An increased posterior tibial slope (PTS) corresponds with an increased risk of graft failure after anterior cruciate ligament (ACL) reconstruction (ACLR). Validated methods of manual PTS measurements are subject to potential interobserver variability and can be inefficient on large datasets. Purpose/Hypothesis: To develop a deep learning artificial intelligence technique for automated PTS measurement from standard lateral knee radiographs. It was hypothesized that this deep learning tool would be able to measure the PTS on a high volume of radiographs expeditiously and that these measurements would be similar to previously validated manual measurements. Study Design: Cohort study (diagnosis); Level of evidence, 2. Methods: A deep learning U-Net model was developed on a cohort of 300 postoperative short-leg lateral radiographs from patients who underwent ACLR to segment the tibial shaft, tibial joint surface, and tibial tuberosity. The model was trained via a random split after an 80 to 20 train-validation scheme. Masks for training images were manually segmented, and the model was trained for 400 epochs. An image processing pipeline was then deployed to annotate and measure the PTS using the predicted segmentation masks. Finally, the performance of this combined pipeline was compared with human measurements performed by 2 study personnel using a previously validated manual technique for measuring the PTS on short-leg lateral radiographs on an independent test set consisting of both pre- and postoperative images. Results: The U-Net semantic segmentation model achieved a mean Dice similarity coefficient of 0.885 on the validation cohort. The mean difference between the human-made and computer-vision measurements was 1.92° (σ = 2.81° [ P = .24]). Extreme disagreements between the human and machine measurements, as defined by ≥5° differences, occurred <5% of the time. The model was incorporated into a web-based digital application front-end for demonstration purposes, which can measure a single uploaded image in Portable Network Graphics format in a mean time of 5 seconds. Conclusion: We developed an efficient and reliable deep learning computer vision algorithm to automate the PTS measurement on short-leg lateral knee radiographs. This tool, which demonstrated good agreement with human annotations, represents an effective clinical adjunct for measuring the PTS as part of the preoperative assessment of patients with ACL injuries.

Introduction: In many cases, treatment decisions for multiple myeloma patients must be made in the absence of high quality randomized controlled clinical trials. As a result, many clinicians lean on prognostic models to drive treatment selection. The most widely used of these models incorporate a limited number of patient related factors that reflect tumor burden, cytogenetic features, or gene expression profiling. These prediction scores are derived from regression models that incorporate these variables at the time of diagnosis. They do not account for variables that change dynamically over time, do not explore the relationship between treatment selection and progression, and imperfectly predict survival outcomes. We designed a prognostic model based on an ensemble machine learning platform to predict progression in NDMM using Extreme Gradient Boosting Machine (XGBoost) combined with accelerated failure time modeling for survival analysis. Methods: We utilized a large retrospective data set containing treatment and response information for 1127 patients with newly diagnosed multiple myeloma (NDMM) treated at the Cleveland Clinic Foundation between 2000 and 2023. Following data preprocessing based on data completeness, 953 patient records were included in our training, testing, and cross validation datasets. For the purposes of our analysis, the initial dataset was randomly split into training (70%), testing (15%) and validation (15%) subsets at the patient level. We also defined the lower and upper bounds for each subset, which is critical due to the right-censored nature of survival data. Hyperparameters for XGBoost were optimized using Bayesian search, minimizing the negative log-likelihood. A random forest method was applied for the imputation of missing data (MissForest). We then trained the XGBoost model using GPU acceleration for enhanced computational efficiency. Finally, a log-rank test on predicted survival times was used to test the model's performance on patients with and without known progression. Results: Our preprocessed data set had 953 patients with a mean age at disease onset of 65 years and a slight male predominance (55%). Approximately 27% of patients harbored high cytogenetic risk disease and 3.5% of the cohort presented with extramedullary disease a diagnosis. At a median follow up of 35 months, 47% of patients had experienced disease progression or death. Induction therapy included an immunomodulatory drug in 40% of patients and proteasome inhibitors in 22%. Frontline autologous stem cell transplantation followed induction in 28% of patients. Median progression free survival in the overall data set was 44 months. These features were consistent across the randomly assigned training, testing, and validation cohorts. Following data preprocessing, 34 independent clinical and genomic variables including selection of first line treatment were assessed as model inputs each available record. Under optimized parameters, the model was trained with a maximum tree depth = 7 and a learning rate ~ 0.22 for the desired output of progression free survival (PFS). We then queried PFS for both the training and validation sets and divided patients into those known to have progressed and those without a progression event. Survival analysis was undertaken using predicted PFS values for patients known to have progressed and those who remained progression free. For the validation data set, our model successfully discerned between progressed and non-progressed cases (log-rank test statistic = 22.07, p &lt; 0.005) (Figure 1). Discussion: We present a machine learning approach based on regularized gradient boosting that accurately discerns between patients who experience progression and those who remain progression free at a median ~35 months of follow-up in a large retrospective data set in patients presenting with NDMM. Further elaboration of our model will allow for the incorporation of large amounts data to predict survival outcomes on the basis of dynamic variables such as depth of response and treatment selection. With the ability to parse outcomes among multiple myeloma patients at high resolution, future clinicians and clinical trialists may be able to overcome limitations in trial design and patient-level therapy selection for both newly diagnosed and relapsed patients.

Abstract GBM is the most common primary malignant brain tumor in adults. O6-methylguanine DNA methyltransferase (MGMT) promoter methylation status is an important prognostic biomarker that predicts the response to temozolomide and guides the treatment decisions. At present, the only reliable way to determine MGMT promoter methylation status is through the analysis of tumor tissues. Given the limitations and complications of histology-based methods, an imaging-based approach for non-invasively MGMT promoter methylation status prediction is beneficial. This study aimed to develop and compare three different deep-learning-based approaches for predicting MGMT promoter methylation status non-invasively. We obtained 576 T2 weighted images (T2WIs) with their corresponding tumor masks and MGMT promoter methylation status from, The Brain Tumor Segmentation (BraTS) 2021 datasets. Dataset was split into five folds at the patient's level stratified by MGMT promoter methylation status to perform a 5-fold cross-validation. We developed three different models: voxel-wise, slice-wise, and whole-brain. For voxel-wise classification, methylated and unmethylated MGMT tumor masks were made into 1 and 2 with 0 background, respectively. We converted each T2WI into 32x32x32 patches. We trained a 3D-Vnet model for tumor segmentation. After inference, we constructed the whole brain volume based on the patch's coordinates. The final prediction of MGMT methylation status was made by majority voting between the predicted voxel values of the biggest connected component. For slice-wise classification, we trained an object detection model for tumor detection and MGMT methylation status prediction; then, we used majority voting for the final prediction. For the whole-brain approach, we trained a 3D Densenet121 for prediction. Whole-brain, slice-wise, voxel-wise, accuracy was 65.42%(SD 3.97%), 61.37%(SD 1.48%), and 56.84%(SD 4.38%) respectively. We found that across the whole-brain, slice-wise, and voxel-wise deep learning approaches, the whole-brain approach is the most effective approach for MGMT methylation status prediction on the BraTS 2021 dataset.

Abstract PURPOSE Though patients with glioblastoma generally have poor overall survival (OS), OS estimates range significantly. We sought to develop a radiomics-based model predictive of OS in glioblastoma using preoperative MRI scans to identify these differences and guide clinical care. METHODS The primary endpoint was OS, determined from the date of surgery to death or last follow-up. T1, FLAIR, T2, and post-contrast T1 (T1c) MRI sequences were obtained for all patients. Using HD-GLIO-AUTO, lesion segmentation yielded three masks corresponding to the regions of contrast enhancement, non-enhancement, and total volume. Radiomics features were extracted from each region from the T2 and T1c sequences independently (i.e. either from T2 or T1c) and in combination. Age at surgery and gender were also considered. The data was split into training (75%) and test (25%) sets, stratifying for age at surgery. Kaplan-Meier analysis and the Coxnet model were used. For Coxnet, Elastic Net was used as the penalty, with optimization of the L1 coefficient using 5-fold cross-validation and evaluation via the concordance index (C-index). RESULTS From 3 institutions, 383 patients were included. Median OS was 395 days: 346 patients died, and 37 were right-censored. Age &gt; 65 years was predictive of decreased OS (p &lt; 0.005), but gender was not (p=0.16). For the Coxnet model, the C-index was 0.67 (standard deviation: 0.01) considering T2 and T1c sequences for the total lesion volume, with a C-index of 0.65 for the test set. For the enhancing volume, the model had a C-index of 0.71 for the test set using T1c sequences. Overall, C-indices ranged from 0.62-0.71. CONCLUSIONS We present a novel radiomics-based model predictive of OS in glioblastoma across a multi-institutional dataset. Ongoing efforts will incorporate immediate postoperative imaging and IDH and MGMT status into the proposed model to further improve prognostic ability and help guide clinical decision-making.

Abstract: Electronic health records (EHRs) have created great opportunities to collect various information from clinical patient encounters. However, most EHR data are stored in unstructured form (e.g., clinical notes, surgical notes, and medication instructions), and researchers need data to be in computable form (structured) to extract meaningful relationships involving variables that can influence patient outcomes. Clinical natural language processing (NLP) is the field of extracting structured data from unstructured text documents in EHRs. Clinical text has several characteristics that mandate the use of special techniques to extract structured information from them compared with generic NLP methods. In this article, we define clinical NLP models, introduce different methods of information extraction from unstructured data using NLP, and describe the basic technical aspects of how deep learning-based NLP models work. We conclude by noting the challenges of working with clinical NLP models and summarizing the general steps needed to launch an NLP project.

Machine-learning (ML) and deep-learning (DL) algorithms are part of a group of modeling algorithms that grasp the hidden patterns in data based on a training process, enabling them to extract complex information from the input data. In the past decade, these algorithms have been increasingly used for image processing, specifically in the medical domain. Cardiothoracic imaging is one of the early adopters of ML/DL research, and the COVID-19 pandemic resulted in more research focus on the feasibility and applications of ML/DL in cardiothoracic imaging. In this scoping review, we systematically searched available peer-reviewed medical literature on cardiothoracic imaging and quantitatively extracted key data elements in order to get a big picture of how ML/DL have been used in the rapidly evolving cardiothoracic imaging field. During this report, we provide insights on different applications of ML/DL and some nuances pertaining to this specific field of research. Finally, we provide general suggestions on how researchers can make their research more than just a proof-of-concept and move toward clinical adoption.