Preoperative CECT-Based Multitask Model Predicts Peritoneal Recurrence and Disease-Free Survival in Advanced Ovarian Cancer: A Multicenter Study

RATIONALE AND OBJECTIVES

Peritoneal recurrence is the predominant pattern of recurrence in advanced ovarian cancer (AOC) and portends a dismal prognosis. Accurate prediction of peritoneal recurrence and disease-free survival (DFS) is crucial to identify patients who might benefit from intensive treatment. We aimed to develop a predictive model for peritoneal recurrence and prognosis in AOC.

METHODS

In this retrospective multi-institution study of 515 patients, an end-to-end multi-task convolutional neural network (MCNN) comprising a segmentation convolutional neural network (CNN) and a classification CNN was developed and tested using preoperative CT images, and MCNN-score was generated to indicate the peritoneal recurrence and DFS status in patients with AOC. We evaluated the accuracy of the model for automatic segmentation and predict prognosis.

RESULTS

The MCNN achieved promising segmentation performances with a mean Dice coefficient of 84.3% (range: 78.8%-87.0%). The MCNN was able to predict peritoneal recurrence in the training (AUC 0.87; 95% CI 0.82-0.90), internal test (0.88; 0.85-0.92), and external test set (0.82; 0.78-0.86). Similarly, MCNN demonstrated consistently high accuracy in predicting recurrence, with an AUC of 0.85; 95% CI 0.82-0.88, 0.83; 95% CI 0.80-0.86, and 0.85; 95% CI 0.83-0.88. For patients with a high MCNN-score of recurrence, it was associated with poorer DFS with P < 0.0001 and hazard ratios of 0.1964 (95% CI: 0.1439-0.2680), 0.3249 (95% CI: 0.1896-0.5565), and 0.3458 (95% CI: 0.2582-0.4632).

CONCLUSION

The MCNN approach demonstrated high performance in predicting peritoneal recurrence and DFS in patients with AOC.

Multicentric development and validation of a multi-scale and multi-task deep learning model for comprehensive lower extremity alignment analysis

Osteoarthritis of the knee, a widespread cause of knee disability, is commonly treated in orthopedics due to its rising prevalence. Lower extremity misalignment, pivotal in knee injury etiology and management, necessitates comprehensive mechanical alignment evaluation via frequently-requested weight-bearing long leg radiographs (LLR). Despite LLR's routine use, current analysis techniques are error-prone and time-consuming. To address this, we conducted a multicentric study to develop and validate a deep learning (DL) model for fully automated leg alignment assessment on anterior-posterior LLR, targeting enhanced reliability and efficiency. The DL model, developed using 594 patients' LLR and a 60%/10%/30% data split for training, validation, and testing, executed alignment analyses via a multi-step process, employing a detection network and nine specialized networks. It was designed to assess all vital anatomical and mechanical parameters for standard clinical leg deformity analysis and preoperative planning. Accuracy, reliability, and assessment duration were compared with three specialized orthopedic surgeons across two distinct institutional datasets (136 and 143 radiographs). The algorithm exhibited equivalent performance to the surgeons in terms of alignment accuracy (DL: 0.21 ± 0.18°to 1.06 ± 1.3°vs. OS: 0.21 ± 0.16°to 1.72 ± 1.96°), interrater reliability (ICC DL: 0.90 ± 0.05 to 1.0 ± 0.0 vs. ICC OS: 0.90 ± 0.03 to 1.0 ± 0.0), and clinically acceptable accuracy (DL: 53.9%-100% vs OS 30.8%-100%). Further, automated analysis significantly reduced analysis time compared to manual annotation (DL: 22 ± 0.6 s vs. OS; 101.7 ± 7 s, p ≤ 0.01). By demonstrating that our algorithm not only matches the precision of expert surgeons but also significantly outpaces them in both speed and consistency of measurements, our research underscores a pivotal advancement in harnessing AI to enhance clinical efficiency and decision-making in orthopaedics.

A multitask approach for automated detection and segmentation of thyroid nodules in ultrasound images

An increase in the incidence and diagnosis of thyroid nodules and thyroid cancer underscores the need for a better approach to nodule detection and risk stratification in ultrasound (US) images that can reduce healthcare costs, patient discomfort, and unnecessary invasive procedures. However, variability in ultrasound technique and interpretation makes the diagnostic process partially subjective. Therefore, an automated approach that detects and segments nodules could improve performance on downstream tasks, such as risk stratification. Ultrasound studies were acquired from 280 patients at UCLA Health, totaling 9888 images, and annotated by collaborating radiologists. Current deep learning architectures for segmentation are typically semi-automated because they are evaluated solely on images known to have nodules and do not assess ability to identify suspicious images. However, the proposed multitask approach both detects suspicious images and segments potential nodules; this allows for a clinically translatable model that aptly parallels the workflow for thyroid nodule assessment. The multitask approach is centered on an anomaly detection (AD) module that can be integrated with any UNet architecture variant to improve image-level nodule detection. Of the evaluated multitask models, a UNet with a ImageNet pretrained encoder and AD achieved the highest F1 score of 0.839 and image-wide Dice similarity coefficient of 0.808 on the hold-out test set. Furthermore, models were evaluated on two external validations datasets to demonstrate generalizability and robustness to data variability. Ultimately, the proposed architecture is an automated multitask method that expands on previous methods by successfully both detecting and segmenting nodules in ultrasound.

PFCE2: A versatile parameter-free calibration enhancement framework for near-infrared spectroscopy

Near-infrared (NIR) spectroscopy is a widely used technique for chemical analysis, but it has faced challenges of calibration transfer, maintenance, and enhancement among different instruments and conditions. The parameter-free calibration enhancement (PFCE) framework was developed to address these challenges with non-supervised (NS), semi-supervised (SS), and full-supervised (FS) methods. This study presented PFCE2, an updated version of the PFCE framework that incorporates two new constraints and a new method to improve the robustness and efficiency of calibration enhancement. First, normalized L2 and L1 constraints were introduced to replace the correlation coefficient (Corr) constraint used in the original PFCE. These constraints preserve the parameter-free feature of PFCE and impose smoothness or sparsity on the model coefficients. Second, multitask PFCE (MT-PFCE) was proposed within the framework to address the calibration enhancement among multiple instruments, enabling the framework to be versatile for all possible calibration transfer situations. Demonstrations conducted on three NIR datasets of tablets, plant leaves, and corn showed that the PFCE methods with the new L2 and L1 constraints can result in more accurate and robust predictions than the Corr constraint, especially when the standard sample size is small. Moreover, MT-PFCE could refine all models in the involved scenarios at once, leading to significant enhancement in model performance, compared to the original PFCE method with the same data requirements. Finally, the applicable situations of the PFCE framework and other analogous calibration transfer methods were summarized, facilitating users to choose suitable methods for their application. The source codes written in both MATLAB and Python are available at https://github.com/JinZhangLab/PFCE and https://pypi.org/project/pynir/, respectively.

Dual-task versus single-task gait rehabilitation after stroke: the protocol of the cognitive-motor synergy multicenter, randomized, controlled superiority trial (SYNCOMOT)

BACKGROUND

Gait disorders and cognitive impairments are prime causes of disability and institutionalization after stroke. We hypothesized that relative to single-task gait rehabilitation (ST GR), cognitive-motor dual-task (DT) GR initiated at the subacute stage would be associated with greater improvements in ST and DT gait, balance, and cognitive performance, personal autonomy, disability, and quality of life in the short, medium and long terms after stroke.

METHODS

This multicenter (n=12), two-arm, parallel-group, randomized (1:1), controlled clinical study is a superiority trial. With p<0.05, a power of 80%, and an expected loss to follow-up rate of 10%, the inclusion of 300 patients will be required to evidence a 0.1-m.s^-1 gain in gait speed. Trial will include adult patients (18-90 years) in the subacute phase (0 to 6 months after a hemispheric stroke) and who are able to walk for 10 m (with or without a technical aid). Registered physiotherapists will deliver a standardized GR program (30 min three times a week, for 4 weeks). The GR program will comprise various DTs (phasic, executive function, praxis, memory, and spatial cognition tasks during gait) in the DT (experimental) group and gait exercises only in the ST (control) group. The primary outcome measure is gait speed 6 months after inclusion. The secondary outcomes are post-stroke impairments (National Institutes of Health Stroke Scale and the motor part of the Fugl-Meyer Assessment of the lower extremity), gait speed (10-m walking test), mobility and dynamic balance (timed up-and-go test), ST and DT cognitive function (the French adaptation of the harmonization standards neuropsychological battery, and eight cognitive-motor DTs), personal autonomy (functional independence measure), restrictions in participation (structured interview and the modified Rankin score), and health-related quality of life (on a visual analog scale). These variables will be assessed immediately after the end of the protocol (probing the short-term effect), 1 month thereafter (the medium-term effect), and 5 months thereafter (the long-term effect).

DISCUSSION

The main study limitation is the open design. The trial will focus on a new GR program applicable at various stages after stroke and during neurological disease.

TRIAL REGISTRATION

NCT03009773 . Registered on January 4, 2017.

A multitask transfer learning framework for the prediction of virus-human protein-protein interactions

BACKGROUND

Viral infections are causing significant morbidity and mortality worldwide. Understanding the interaction patterns between a particular virus and human proteins plays a crucial role in unveiling the underlying mechanism of viral infection and pathogenesis. This could further help in prevention and treatment of virus-related diseases. However, the task of predicting protein-protein interactions between a new virus and human cells is extremely challenging due to scarce data on virus-human interactions and fast mutation rates of most viruses.

RESULTS

We developed a multitask transfer learning approach that exploits the information of around 24 million protein sequences and the interaction patterns from the human interactome to counter the problem of small training datasets. Instead of using hand-crafted protein features, we utilize statistically rich protein representations learned by a deep language modeling approach from a massive source of protein sequences. Additionally, we employ an additional objective which aims to maximize the probability of observing human protein-protein interactions. This additional task objective acts as a regularizer and also allows to incorporate domain knowledge to inform the virus-human protein-protein interaction prediction model.

CONCLUSIONS

Our approach achieved competitive results on 13 benchmark datasets and the case study for the SARS-COV-2 virus receptor. Experimental results show that our proposed model works effectively for both virus-human and bacteria-human protein-protein interaction prediction tasks. We share our code for reproducibility and future research at https://git.l3s.uni-hannover.de/dong/multitask-transfer .