Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

oncology Oncology

A hybrid model- and deep learning-based framework for functional lung image synthesis from multi-inflation CT and hyperpolarized gas MRI.

In Medical physics ; h5-index 59.0

BACKGROUND : Hyperpolarized gas MRI is a functional lung imaging modality capable of visualizing regional lung ventilation with exceptional detail within a single breath. However, this modality requires specialized equipment and exogenous contrast, which limits widespread clinical adoption. CT ventilation imaging employs various metrics to model regional ventilation from non-contrast CT scans acquired at multiple inflation levels and has demonstrated moderate spatial correlation with hyperpolarized gas MRI. Recently, deep learning (DL)-based methods, utilizing convolutional neural networks (CNNs), have been leveraged for image synthesis applications. Hybrid approaches integrating computational modeling and data-driven methods have been utilized in cases where datasets are limited with the added benefit of maintaining physiological plausibility.

PURPOSE : To develop and evaluate a multi-channel DL-based method that combines modeling and data-driven approaches to synthesize hyperpolarized gas MRI lung ventilation scans from multi-inflation, non-contrast CT and quantitatively compare these synthetic ventilation scans to conventional CT ventilation modeling.

METHODS : In this study, we propose a hybrid DL configuration that integrates model- and data-driven methods to synthesize hyperpolarized gas MRI lung ventilation scans from a combination of non-contrast, multi-inflation CT and CT ventilation modeling. We used a diverse dataset comprising paired inspiratory and expiratory CT and helium-3 hyperpolarized gas MRI for 47 participants with a range of pulmonary pathologies. We performed 6-fold cross-validation on the dataset and evaluated the spatial correlation between the synthetic ventilation and real hyperpolarized gas MRI scans; the proposed hybrid framework was compared to conventional CT ventilation modeling and other non-hybrid DL configurations. Synthetic ventilation scans were evaluated using voxel-wise evaluation metrics such as Spearman's correlation and mean square error (MSE), in addition to clinical biomarkers of lung function such as the ventilated lung percentage (VLP). Furthermore, regional localization of ventilated and defect lung regions was assessed via the Dice similarity coefficient (DSC).

RESULTS : We showed that the proposed hybrid framework is capable of accurately replicating ventilation defects seen in the real hyperpolarized gas MRI scans, achieving a voxel-wise Spearman's correlation of 0.57±0.17 and an MSE of 0.017±0.01. The hybrid framework significantly outperformed CT ventilation modeling alone and all other DL configurations using Spearman's correlation. The proposed framework was capable of generating clinically relevant metrics such as the VLP without manual intervention, resulting in a Bland-Altman bias of 3.04%, significantly outperforming CT ventilation modeling. Relative to CT ventilation modeling, the hybrid framework yielded significantly more accurate delineations of ventilated and defect lung regions, achieving a DSC of 0.95 and 0.48 for ventilated and defect regions, respectively.

CONCLUSIONS : The ability to generate realistic synthetic ventilation scans from CT has implications for several clinical applications, including functional lung avoidance radiotherapy and treatment response mapping. CT is an integral part of almost every clinical lung imaging workflow and hence is readily available for most patients; therefore, synthetic ventilation from non-contrast CT can provide patients with wider access to ventilation imaging worldwide. This article is protected by copyright. All rights reserved.

Astley Joshua R, Biancardi Alberto M, Marshall Helen, Hughes Paul J C, Collier Guilhem J, Hatton Matthew Q, Wild Jim M, Tahir Bilal A

2023-Mar-17

18 Claremont crescent, POLARIS, S10 2TA, Sheffield, United Kingdom

General General

Multi-task adaptive pooling enabled synergetic learning of RNA modification across tissue, type and species from low-resolution epitranscriptomes.

In Briefings in bioinformatics

Post- and co-transcriptional RNA modifications are found to play various roles in regulating essential biological processes at all stages of RNA life. Precise identification of RNA modification sites is thus crucial for understanding the related molecular functions and specific regulatory circuitry. To date, a number of computational approaches have been developed for in silico identification of RNA modification sites; however, most of them require learning from base-resolution epitranscriptome datasets, which are generally scarce and available only for a limited number of experimental conditions, and predict only a single modification, even though there are multiple inter-related RNA modification types available. In this study, we proposed AdaptRM, a multi-task computational method for synergetic learning of multi-tissue, type and species RNA modifications from both high- and low-resolution epitranscriptome datasets. By taking advantage of adaptive pooling and multi-task learning, the newly proposed AdaptRM approach outperformed the state-of-the-art computational models (WeakRM and TS-m6A-DL) and two other deep-learning architectures based on Transformer and ConvMixer in three different case studies for both high-resolution and low-resolution prediction tasks, demonstrating its effectiveness and generalization ability. In addition, by interpreting the learned models, we unveiled for the first time the potential association between different tissues in terms of epitranscriptome sequence patterns. AdaptRM is available as a user-friendly web server from http://www.rnamd.org/AdaptRM together with all the codes and data used in this project.

Song Yiyou, Wang Yue, Wang Xuan, Huang Daiyun, Nguyen Anh, Meng Jia

2023-Mar-17

ConvMixer, RNA modification, adaptive pooling, low-resolution epitranscriptomes, multi-task learning, transformer

General General

Identification of FCN1 as a novel macrophage infiltration-associated biomarker for diagnosis of pediatric inflammatory bowel diseases.

In Journal of translational medicine

BACKGROUND : The incidence of pediatric inflammatory bowel disease (PIBD) has been steadily increasing globally. Delayed diagnosis of PIBD increases the risk of complications and contributes to growth retardation. To improve long-term outcomes, there is a pressing need to identify novel markers for early diagnosis of PIBD.

METHODS : The candidate biomarkers for PIBD were identified from the GSE117993 dataset by two machine learning algorithms, namely LASSO and mSVM-RFE, and externally validated in the GSE126124 dataset and our PIBD cohort. The role of ficolin-1 (FCN1) in PIBD and its association with macrophage infiltration was investigated using the CIBERSORT method and enrichment analysis of the single-cell dataset GSE121380, and further validated using immunoblotting, qRT-PCR, and immunostaining in colon biopsies from PIBD patients, a juvenile murine DSS-induced colitis model, and THP-1-derived macrophages.

RESULTS : FCN1 showed great diagnostic performance for PIBD in an independent clinical cohort with the AUC of 0.986. FCN1 expression was upregulated in both colorectal biopsies and blood samples from PIBD patients. Functionally, FCN1 was associated with immune-related processes in the colonic mucosa of PIBD patients, and correlated with increased proinflammatory M1 macrophage infiltration. Furthermore, single-cell transcriptome analysis and immunostaining revealed that FCN1 was almost exclusively expressed in macrophages infiltrating the colonic mucosa of PIBD patients, and these FCN1+ macrophages were related to hyper-inflammation. Notably, proinflammatory M1 macrophages derived from THP-1 expressed high levels of FCN1 and IL-1β, and FCN1 overexpression in THP-1-derived macrophages strongly promoted LPS-induced activation of the proinflammatory cytokine IL-1β via the NLRP3-caspase-1 axis.

CONCLUSIONS : FCN1 is a novel and promising diagnostic biomarker for PIBD. FCN1+ macrophages enriched in the colonic mucosa of PIBD exhibit proinflammatory phenotypes, and FCN1 promotes IL-1β maturation in macrophages via the NLRP3-caspase-1 axis.

Chen Xixi, Gao Yuanqi, Xie Jinfang, Hua Huiying, Pan Chun, Huang Jiebin, Jing Mengxia, Chen Xuehua, Xu Chundi, Gao Yujing, Li Pu

2023-Mar-17

Diagnostic biomarkers, FCN1, Inflammation, Macrophages, Pediatric inflammatory bowel disease

Radiology Radiology

Effect of Dataset Size and Medical Image Modality on Convolutional Neural Network Model Performance for Automated Segmentation: A CT and MR Renal Tumor Imaging Study.

In Journal of digital imaging

The aim of this study is to investigate the use of an exponential-plateau model to determine the required training dataset size that yields the maximum medical image segmentation performance. CT and MR images of patients with renal tumors acquired between 1997 and 2017 were retrospectively collected from our nephrectomy registry. Modality-based datasets of 50, 100, 150, 200, 250, and 300 images were assembled to train models with an 80-20 training-validation split evaluated against 50 randomly held out test set images. A third experiment using the KiTS21 dataset was also used to explore the effects of different model architectures. Exponential-plateau models were used to establish the relationship of dataset size to model generalizability performance. For segmenting non-neoplastic kidney regions on CT and MR imaging, our model yielded test Dice score plateaus of [Formula: see text] and [Formula: see text] with the number of training-validation images needed to reach the plateaus of 54 and 122, respectively. For segmenting CT and MR tumor regions, we modeled a test Dice score plateau of [Formula: see text] and [Formula: see text], with 125 and 389 training-validation images needed to reach the plateaus. For the KiTS21 dataset, the best Dice score plateaus for nn-UNet 2D and 3D architectures were [Formula: see text] and [Formula: see text] with number to reach performance plateau of 177 and 440. Our research validates that differing imaging modalities, target structures, and model architectures all affect the amount of training images required to reach a performance plateau. The modeling approach we developed will help future researchers determine for their experiments when additional training-validation images will likely not further improve model performance.

Gottlich Harrison C, Gregory Adriana V, Sharma Vidit, Khanna Abhinav, Moustafa Amr U, Lohse Christine M, Potretzke Theodora A, Korfiatis Panagiotis, Potretzke Aaron M, Denic Aleksandar, Rule Andrew D, Takahashi Naoki, Erickson Bradley J, Leibovich Bradley C, Kline Timothy L

2023-Mar-17

Deep learning, Kidney, Machine learning model performance, Nephrectomy, Semantic segmentation, Similarity metrics

Surgery Surgery

Surgery or comfort care for neonates with surgical necrotizing enterocolitis: Lessons learned from behavioral artificial intelligence technology.

In Frontiers in pediatrics

BACKGROUND : Critical decision making in surgical necrotizing enterocolitis (NEC) is highly complex and hard to capture in decision rules due to case-specificity and high mortality risk. In this choice experiment, we aimed to identify the implicit weight of decision factors towards future decision support, and to assess potential differences between specialties or centers.

METHODS : Thirty-five hypothetical surgical NEC scenarios with different factor levels were evaluated by neonatal care experts of all Dutch neonatal care centers in an online environment, where a recommendation for surgery or comfort care was requested. We conducted choice analysis by constructing a binary logistic regression model according to behavioral artificial intelligence technology (BAIT).

RESULTS : Out of 109 invited neonatal care experts, 62 (57%) participated, including 45 neonatologists, 16 pediatric surgeons and one neonatology physician assistant. Cerebral ultrasound (Relative importance = 20%, OR = 4.06, 95% CI = 3.39-4.86) was the most important factor in the decision surgery versus comfort care in surgical NEC, nationwide and for all specialties and centers. Pediatric surgeons more often recommended surgery compared to neonatologists (62% vs. 57%, p = 0.03). For all centers, cerebral ultrasound, congenital comorbidity, hemodynamics and parental preferences were significant decision factors (p < 0.05). Sex (p = 0.14), growth since birth (p = 0.25), and estimated parental capacities (p = 0.06) had no significance in nationwide nor subgroup analyses.

CONCLUSION : We demonstrated how BAIT can analyze the implicit weight of factors in the complex and critical decision for surgery or comfort care for (surgical) NEC. The findings reflect Dutch expertise, but the technique can be expanded internationally. After validation, our choice model/BAIT may function as decision aid.

van Varsseveld Otis C, Ten Broeke Annebel, Chorus Caspar G, Heyning Nicolaas, Kooi Elisabeth M W, Hulscher Jan B F

2023

artificial intelligence, choice analysis, comfort care, critical care, decision making, decision support, necrotizing enterocolitis

Surgery Surgery

Non-invasive estimation of muscle fibre size from high-density electromyography.

In The Journal of physiology ; h5-index 67.0

Because of the biophysical relation between muscle fibre diameter and the propagation velocity of action potentials along the muscle fibres, motor unit conduction velocity could be a non-invasive index of muscle fibre size in humans. However, the relation between motor unit conduction velocity and fibre size has been only assessed indirectly in animal models and in human patients with invasive intramuscular EMG recordings, or it has been mathematically derived from computer simulations. By combining advanced non-invasive techniques to record motor unit activity in vivo, i.e., high-density surface EMG, with the gold standard technique for muscle tissue sampling, i.e., muscle biopsy, here we investigated the relation between the conduction velocity of populations of motor units identified from the biceps brachii muscle, and muscle fibre diameter. We demonstrate the possibility to predict muscle fibre diameter (R2 = 0.66) and cross-sectional area (R2 = 0.65) from conduction velocity estimates with low systematic bias (∼2% and ∼4% respectively) and a relatively low margin of individual error (∼8% and ∼16%, respectively). The proposed neuromuscular interface opens new perspectives in the use of high-density EMG as a non-invasive tool to estimate muscle fibre size without the need of surgical biopsy sampling. The non-invasive nature of high-density surface EMG for the assessment of muscle fibre size may be useful in studies monitoring child development, aging, space and exercise physiology, although the applicability and validity of the proposed methodology needs to be more directly assessed in these specific populations by future studies. KEY POINTS: Because of the biophysical relation between muscle fibre size and the propagation velocity of action potentials along the sarcolemma, motor unit conduction velocity could represent a potential non-invasive candidate to estimate muscle fibre size in vivo. This relation has been previously assessed in animal models and humans with invasive techniques, or it has been mathematically-derived from simulations. By combining high-density surface EMG with muscle biopsy, here we explored the relation between the conduction velocity of populations of motor units and muscle fibre size in healthy individuals. Our results confirmed that motor unit conduction velocity can be considered as a novel biomarker of fibre size, which can be adopted to predict muscle fibre diameter and cross-sectional area with low systematic bias and margin of individual error. The proposed neuromuscular interface opens new perspectives in the use of high-density EMG as a non-invasive tool to estimate muscle fibre size without the need of surgical biopsy sampling Abstract figure legend In this study, we investigated the relation between the conduction velocity of populations of motor units identified from biceps brachii muscle and muscle fibre size. We adopted high-density surface EMG to decode the activity of voluntarily activated motor units and estimated their conduction velocity. Similarly, we adopted muscle biopsy to measure muscle fibre size. We revealed the possibility to accurately transform motor unit conduction velocity values into estimated measures of muscle fibre size, which in turn showed a good degree of association with the muscle fibre size measured directly by muscle biopsy. Furthermore, we demonstrated that the proposed neuromuscular interface allows to predict the mean measured fibre diameter and cross-sectional area from an EMG-derived parameter with a relatively low bias and error, thus opening new perspectives in the use of high-density EMG as a non-invasive tool to estimate muscle fibre size without the need of surgical biopsy sampling This article is protected by copyright. All rights reserved.

Casolo Andrea, Maeo Sumiaki, Balshaw Thomas G, Lanza Marcel B, Martin Neil R W, Nuccio Stefano, Moro Tatiana, Paoli Antonio, Felici Francesco, Maffulli Nicola, Eskofier Bjoern, Kinfe Thomas M, Folland Jonathan P, Farina Dario, Vecchio Alessandro Del

2023-Mar-16

Motor unit, conduction velocity, high-density surface electromyography, muscle fibre size