skip to main content

Title: Siamese neural networks for continuous disease severity evaluation and change detection in medical imaging

Using medical images to evaluate disease severity and change over time is a routine and important task in clinical decision making. Grading systems are often used, but are unreliable as domain experts disagree on disease severity category thresholds. These discrete categories also do not reflect the underlying continuous spectrum of disease severity. To address these issues, we developed a convolutional Siamese neural network approach to evaluate disease severity at single time points and change between longitudinal patient visits on a continuous spectrum. We demonstrate this in two medical imaging domains: retinopathy of prematurity (ROP) in retinal photographs and osteoarthritis in knee radiographs. Our patient cohorts consist of 4861 images from 870 patients in the Imaging and Informatics in Retinopathy of Prematurity (i-ROP) cohort study and 10,012 images from 3021 patients in the Multicenter Osteoarthritis Study (MOST), both of which feature longitudinal imaging data. Multiple expert clinician raters ranked 100 retinal images and 100 knee radiographs from excluded test sets for severity of ROP and osteoarthritis, respectively. The Siamese neural network output for each image in comparison to a pool of normal reference images correlates with disease severity rank (ρ = 0.87 for ROP andρ = 0.89 for osteoarthritis), both within and between the clinical grading categories. Thus, this output can represent the continuous spectrum of disease severity at any single time point. The difference in these outputs can be used to show change over time. Alternatively, paired images from the same patient at two time points can be directly compared using the Siamese neural network, resulting in an additional continuous measure of change between images. Importantly, our approach does not require manual localization of the pathology of interest and requires only a binary label for training (same versus different). The location of disease and site of change detected by the algorithm can be visualized using an occlusion sensitivity map-based approach. For a longitudinal binary change detection task, our Siamese neural networks achieve test set receiving operator characteristic area under the curves (AUCs) of up to 0.90 in evaluating ROP or knee osteoarthritis change, depending on the change detection strategy. The overall performance on this binary task is similar compared to a conventional convolutional deep-neural network trained for multi-class classification. Our results demonstrate that convolutional Siamese neural networks can be a powerful tool for evaluating the continuous spectrum of disease severity and change in medical imaging.

more » « less
Author(s) / Creator(s):
; ; ; ; ; ; ; ; ; ; ; ; ;
Publisher / Repository:
Nature Publishing Group
Date Published:
Journal Name:
npj Digital Medicine
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. null (Ed.)
    Osteoarthritis (OA) is the most common form of arthritis and can often occur in the knee. While convolutional neural networks (CNNs) have been widely used to study medical images, the application of a 3-dimensional (3D) CNN in knee OA diagnosis is limited. This study utilizes a 3D CNN model to analyze sequences of knee magnetic resonance (MR) images to perform knee OA classification. An advantage of using 3D CNNs is the ability to analyze the whole sequence of 3D MR images as a single unit as opposed to a traditional 2D CNN, which examines one image at a time. Therefore, 3D features could be extracted from adjacent slices, which may not be detectable from a single 2D image. The input data for each knee were a sequence of double-echo steady-state (DESS) MR images, and each knee was labeled by the Kellgren and Lawrence (KL) grade of severity at levels 0–4. In addition to the 5-category KL grade classification, we further examined a 2-category classification that distinguishes non-OA (KL ≤ 1) from OA (KL ≥ 2) knees. Clinically, diagnosing a patient with knee OA is the ultimate goal of assigning a KL grade. On a dataset with 1100 knees, the 3D CNN model that classifies knees with and without OA achieved an accuracy of 86.5% on the validation set and 83.0% on the testing set. We further conducted a comparative study between MRI and X-ray. Compared with a CNN model using X-ray images trained from the same group of patients, the proposed 3D model with MR images achieved higher accuracy in both the 5-category classification (54.0% vs. 50.0%) and the 2-category classification (83.0% vs. 77.0%). The result indicates that MRI, with the application of a 3D CNN model, has greater potential to improve diagnosis accuracy for knee OA clinically than the currently used X-ray methods. 
    more » « less
  2. In the medical sector, three-dimensional (3D) images are commonly used like computed tomography (CT) and magnetic resonance imaging (MRI). The 3D MRI is a non-invasive method of studying the soft-tissue structures in a knee joint for osteoarthritis studies. It can greatly improve the accuracy of segmenting structures such as cartilage, bone marrow lesion, and meniscus by identifying the bone structure first. U-net is a convolutional neural network that was originally designed to segment the biological images with limited training data. The input of the original U-net is a single 2D image and the output is a binary 2D image. In this study, we modified the U-net model to identify the knee bone structures using 3D MRI, which is a sequence of 2D slices. A fully automatic model has been proposed to detect and segment knee bones. The proposed model was trained, tested, and validated using 99 knee MRI cases where each case consists of 160 2D slices for a single knee scan. To evaluate the model’s performance, the similarity, dice coefficient (DICE), and area error metrics were calculated. Separate models were trained using different knee bone components including tibia, femur, patella, as well as a combined model for segmenting all the knee bones. Using the whole MRI sequence (160 slices), the method was able to detect the beginning and ending bone slices first, and then segment the bone structures for all the slices in between. On the testing set, the detection model accomplished 98.79% accuracy and the segmentation model achieved DICE 96.94% and similarity 93.98%. The proposed method outperforms several state-of-the-art methods, i.e., it outperforms U-net by 3.68%, SegNet by 14.45%, and FCN-8 by 2.34%, in terms of DICE score using the same dataset. 
    more » « less
  3. In many mechanistic medical, biological, physical, and engineered spatiotemporal dynamic models the numerical solution of partial differential equations (PDEs), especially for diffusion, fluid flow and mechanical relaxation, can make simulations impractically slow. Biological models of tissues and organs often require the simultaneous calculation of the spatial variation of concentration of dozens of diffusing chemical species. One clinical example where rapid calculation of a diffusing field is of use is the estimation of oxygen gradients in the retina, based on imaging of the retinal vasculature, to guide surgical interventions in diabetic retinopathy. Furthermore, the ability to predict blood perfusion and oxygenation may one day guide clinical interventions in diverse settings, i.e., from stent placement in treating heart disease to BOLD fMRI interpretation in evaluating cognitive function (Xie et al., 2019 ; Lee et al., 2020 ). Since the quasi-steady-state solutions required for fast-diffusing chemical species like oxygen are particularly computationally costly, we consider the use of a neural network to provide an approximate solution to the steady-state diffusion equation. Machine learning surrogates, neural networks trained to provide approximate solutions to such complicated numerical problems, can often provide speed-ups of several orders of magnitude compared to direct calculation. Surrogates of PDEs could enable use of larger and more detailed models than are possible with direct calculation and can make including such simulations in real-time or near-real time workflows practical. Creating a surrogate requires running the direct calculation tens of thousands of times to generate training data and then training the neural network, both of which are computationally expensive. Often the practical applications of such models require thousands to millions of replica simulations, for example for parameter identification and uncertainty quantification, each of which gains speed from surrogate use and rapidly recovers the up-front costs of surrogate generation. We use a Convolutional Neural Network to approximate the stationary solution to the diffusion equation in the case of two equal-diameter, circular, constant-value sources located at random positions in a two-dimensional square domain with absorbing boundary conditions. Such a configuration caricatures the chemical concentration field of a fast-diffusing species like oxygen in a tissue with two parallel blood vessels in a cross section perpendicular to the two blood vessels. To improve convergence during training, we apply a training approach that uses roll-back to reject stochastic changes to the network that increase the loss function. The trained neural network approximation is about 1000 times faster than the direct calculation for individual replicas. Because different applications will have different criteria for acceptable approximation accuracy, we discuss a variety of loss functions and accuracy estimators that can help select the best network for a particular application. We briefly discuss some of the issues we encountered with overfitting, mismapping of the field values and the geometrical conditions that lead to large absolute and relative errors in the approximate solution. 
    more » « less
  4. Clinical outcome or severity prediction from medical images has largely focused on learning representations from single-timepoint or snapshot scans. It has been shown that disease progression can be better characterized by temporal imaging. We therefore hypothesized that outcome predictions can be improved by utilizing the disease progression information from sequential images. We present a deep learning approach that leverages temporal progression information to improve clinical outcome predictions from single-timepoint images. In our method, a self-attention based Temporal Convolutional Network (TCN) is used to learn a representation that is most reflective of the disease trajectory. Meanwhile, a Vision Transformer is pretrained in a self-supervised fashion to extract features from single-timepoint images. The key contribution is to design a recalibration module that employs maximum mean discrepancy loss (MMD) to align distributions of the above two contextual representations. We train our system to predict clinical outcomes and severity grades from single-timepoint images. Experiments on chest and osteoarthritis radiography datasets demonstrate that our approach outperforms other state-of-the-art techniques. 
    more » « less
  5. Abstract Background We aimed to determine if composite structural measures of knee osteoarthritis (KOA) progression on magnetic resonance (MR) imaging can predict the radiographic onset of accelerated knee osteoarthritis. Methods We used data from a nested case-control study among participants from the Osteoarthritis Initiative without radiographic KOA at baseline. Participants were separated into three groups based on radiographic disease progression over 4 years: 1) accelerated (Kellgren-Lawrence grades [KL] 0/1 to 3/4), 2) typical (increase in KL, excluding accelerated osteoarthritis), or 3) no KOA (no change in KL). We assessed tibiofemoral cartilage damage (four regions: medial/lateral tibia/femur), bone marrow lesion (BML) volume (four regions: medial/lateral tibia/femur), and whole knee effusion-synovitis volume on 3 T MR images with semi-automated programs. We calculated two MR-based composite scores. Cumulative damage was the sum of standardized cartilage damage. Disease activity was the sum of standardized volumes of effusion-synovitis and BMLs. We focused on annual images from 2 years before to 2 years after radiographic onset (or a matched time for those without knee osteoarthritis). To determine between group differences in the composite metrics at all time points, we used generalized linear mixed models with group (3 levels) and time (up to 5 levels). For our prognostic analysis, we used multinomial logistic regression models to determine if one-year worsening in each composite metric change associated with future accelerated knee osteoarthritis (odds ratios [OR] based on units of 1 standard deviation of change). Results Prior to disease onset, the accelerated KOA group had greater average disease activity compared to the typical and no KOA groups and this persisted up to 2 years after disease onset. During a pre-radiographic disease period, the odds of developing accelerated KOA were greater in people with worsening disease activity [versus typical KOA OR (95% confidence interval [CI]): 1.58 (1.08 to 2.33); versus no KOA: 2.39 (1.55 to 3.71)] or cumulative damage [versus typical KOA: 1.69 (1.14 to 2.51); versus no KOA: 2.11 (1.41 to 3.16)]. Conclusions MR-based disease activity and cumulative damage metrics may be prognostic markers to help identify people at risk for accelerated onset and progression of knee osteoarthritis. 
    more » « less