Biologically interpretable multi-task deep learning pipeline predicts molecular alterations, grade, and prognosis in glioma patients.

Deep learning models have been developed for various predictions in glioma; yet, they were constrained by manual segmentation, task-specific design, or a lack of biological interpretation. Herein, we aimed to develop an end-to-end multi-task deep learning (MDL) pipeline that can simultaneously predict molecular alterations and histological grade (auxiliary tasks), as well as prognosis (primary task) in gliomas. Further, we aimed to provide the biological mechanisms underlying the model's predictions. We collected multiscale data including baseline MRI images from 2776 glioma patients across two private (FAHZU and HPPH, n = 1931) and three public datasets (TCGA, n = 213; UCSF, n = 410; and EGD, n = 222). We trained and internally validated the MDL model using our private datasets, and externally validated it using the three public datasets. We used the model-predicted deep prognosis score (DPS) to stratify patients into low-DPS and high-DPS subtypes. Additionally, a radio-multiomics analysis was conducted to elucidate the biological basis of the DPS. In the external validation cohorts, the MDL model achieved average areas under the curve of 0.892-0.903, 0.710-0.894, and 0.850-0.879 for predicting IDH mutation status, 1p/19q co-deletion status, and tumor grade, respectively. Moreover, the MDL model yielded a C-index of 0.723 in the TCGA and 0.671 in the UCSF for the prediction of overall survival. The DPS exhibits significant correlations with activated oncogenic pathways, immune infiltration patterns, specific protein expression, DNA methylation, tumor mutation burden, and tumor-stroma ratio. Accordingly, our work presents an accurate and biologically meaningful tool for predicting molecular subtypes, tumor grade, and survival outcomes in gliomas, which provides personalized clinical decision-making in a global and non-invasive manner.

An interpretable machine learning model based on contrast-enhanced CT parameters for predicting treatment response to conventional transarterial chemoembolization in patients with hepatocellular carcinoma.

OBJECTIVE: To explore the potential of pre-therapy computed tomography (CT) parameters in predicting the treatment response to initial conventional TACE (cTACE) in intermediate-stage hepatocellular carcinoma (HCC) and develop an interpretable machine learning model. METHODS: This retrospective study included 367 patients with intermediate-stage HCC who received cTACE as first-line therapy from three centers. We measured the mean attenuation values of target lesions on multi-phase contrast-enhanced CT and further calculated three CT parameters, including arterial (AER), portal venous (PER), and arterial portal venous (APR) enhancement ratios. We used logistic regression analysis to select discriminative features and trained three machine learning models via 5-fold cross-validation. The performance in predicting treatment response was evaluated in terms of discrimination, calibration, and clinical utility. Afterward, a Shapley additive explanation (SHAP) algorithm was leveraged to interpret the outputs of the best-performing model. RESULTS: The mean diameter, ECOG performance status, and cirrhosis were the important clinical predictors of cTACE treatment response, by multiple logistic regression. Adding the CT parameters to clinical variables showed significant improvement in performance (net reclassification index, 0.318, P < 0.001). The Random Forest model (hereafter, RF-combined model) integrating CT parameters and clinical variables demonstrated the highest performance on external validation dataset (AUC of 0.800). The decision curve analysis illustrated the optimal clinical benefits of RF-combined model. This model could successfully stratify patients into responders and non-responders with distinct survival (P = 0.001). CONCLUSION: The RF-combined model can serve as a robust and interpretable tool to identify the appropriate crowd for cTACE sessions, sparing patients from receiving ineffective and unnecessary treatments.

The Dual-Targeted Peptide Conjugated Probe for Depicting Residual Nasopharyngeal Carcinoma and Guiding Surgery.

Detecting residual nasopharyngeal carcinoma (rNPC) can be difficult because of the coexistence of occult tumours and post-chemoradiation changes, which poses a challenge for both radiologists and surgeons using current imaging methods. Currently, molecular imaging that precisely targets and visualises particular biomarkers in tumours may exceed the specificity and sensitivity of traditional imaging techniques, providing the potential to distinguish tumours from non-neoplastic lesions. Here, we synthesised a HER2/SR-BI-targeted tracer to efficiently position NPC and guide surgery in living mice. This bispecific tracer contained the following two parts: IRDye 800 CW, as an imaging reagent for both optical and optoacoustic imaging, and a fusion peptide (FY-35), as the targeting reagent. Both in vitro and in vivo tests demonstrated that the tracer had higher accumulation and longer retention (up to 48 h) in tumours than a single-targeted probe, and realised sensitive detection of tumours with a minimum size of 3.9 mm. By visualising the vascular network via a customised handheld optoacoustic scan, our intraoperative fluorescence molecular imaging system provides accurate guidance for intraoperative tumour resection. Integrating the advantages of both optical and optoacoustic scanning in an intraoperative image-guided system, this method holds promise for depicting rNPC and guiding salvage surgery.

Carbonic anhydrase IX stratifies patient prognosis and identifies nodal status in animal models of nasopharyngeal carcinoma using a targeted imaging strategy.

PURPOSE: Accurate identification of nodal status enables adequate neck irradiation for nasopharyngeal carcinoma (NPC). However, most conventional techniques are unable to pick up occult metastases, leading to underestimation of tumor extensions. Here we investigate the clinical significance of carbonic anhydrase IX (CAIX) in human NPC samples, and develop a CAIX-targeted imaging strategy to identify occult lymph node metastases (LNMs) and extranodal extension (ENE) in animal studies. METHODS: A total of 211 NPC samples are performed CAIX staining, and clinical outcomes are analyzed. The metastatic murine models are generated by foot pad injection of NPC cells, and a CAIX-targeted imaging agent (CAIX-800) is intravenously administered. We adopt fluorescence molecular tomography and ultrasonography (US)-guided spectroscopic photoacoustic (sPA) imaging to perform in vivo studies. Histological and immunohistochemical characterization are carried out via node-by-node analysis. RESULTS: For clinical samples, 90.1% (91/101) primary tumors, 73.3% (66/90) metastases, and 100% (20/20) local recurrences are CAIX positive. In metastases group, 84.7% (61/72) nodal metastases and 22.2% (4/18) organ metastases are CAIX positive. CAIX expression in primary tumors is significantly associated with NPC stage and prognosis. For animal studies, CAIX-800-based fluorescence imaging achieves 81.3% sensitivity and 93.8% specificity in detecting occult LNMs in vivo, with a minimum detectable diameter of 1.7 mm. Coupled with CAIX-800, US-guided sPA imaging could not only detect subcapsular deposits of metastatic cancer cells 2 weeks earlier than conventional techniques, but also successfully track pathological ENE. CONCLUSION: CAIX remarkably expresses in human NPCs and stratifies patient prognosis. In preclinical studies, CAIX-800-based imaging successfully identifies occult LNMs and tracks early stage of pathological ENE. This attractive method shows potential in clinic, allowing medical workers to longitudinally monitor nodal status and helping to reduce unnecessary nodal biopsy for patients with NPC. The schematic diagram for the study. CAIX, carbonic anhydrase IX; NPC, nasopharyngeal carcinoma; US, ultrasonography; sPA, spectroscopic photoacoustic.

Harnessing the 2D Structure-Enabled Viscoelasticity of Graphene-Based Hydrogel Membranes for Chronic Neural Interfacing.

Stiffness and viscoelasticity of neural implants regulate the foreign body response. Recent studies have suggested the use of elastic or viscoelastic materials with tissue-like stiffness for long-term neural electrical interfacing. Herein, the authors find that a viscoelastic multilayered graphene hydrogel (MGH) membrane, despite exhibiting a much higher Young's modulus than nerve tissues, shows little inflammatory response after 8-week implantation in rat sciatic nerves. The MGH membrane shows significant viscoelasticity due to the slippage between graphene nanosheets, facilitating its seamless yet minimally compressive interfacing with nerves to reduce the inflammation caused by the stiffness mismatch. When used as neural stimulation electrodes, the MGH membrane can offer abundant ion-accessible surfaces to bring a charge injection capacity 1-2 orders of magnitude higher than its traditional Pt counterpart, and further demonstrates chronic neural therapy potential in low-voltage modulation of rat blood pressure. This work suggests that the emergence of 2D nanomaterials and particularly their unique structural attributes can be harnessed to enable new bio-interfacing design strategies.

Tumor-Associated Macrophages and Their Functional Transformation in the Hypoxic Tumor Microenvironment.

Tumor-associated macrophages (TAMs) are some of the most abundant immune cells within tumors and perform a broad repertoire of functions via diverse phenotypes. On the basis of their functional differences in tumor growth, TAMs are usually categorized into two subsets of M1 and M2. It is well established that the tumor microenvironment (TME) is characterized by hypoxia along with tumor progression. TAMs adopt an M1-like pro-inflammatory phenotype at the early phases of oncogenesis and mediate immune response that inhibits tumor growth. As tumors progress, anabatic hypoxia of the TME gradually induces the M2-like functional transformation of TAMs by means of direct effects, metabolic influence, lactic acidosis, angiogenesis, remodeled stroma, and then urges them to participate in immunosuppression, angiogenesis and other tumor-supporting procedure. Therefore, thorough comprehension of internal mechanism of this TAM functional transformation in the hypoxic TME is of the essence, and might provide some novel insights in hypoxic tumor immunotherapeutic strategies.

Domain Transform Network for Photoacoustic Tomography from Limited-view and Sparsely Sampled Data.

Medical image reconstruction methods based on deep learning have recently demonstrated powerful performance in photoacoustic tomography (PAT) from limited-view and sparse data. However, because most of these methods must utilize conventional linear reconstruction methods to implement signal-to-image transformations, their performance is restricted. In this paper, we propose a novel deep learning reconstruction approach that integrates appropriate data pre-processing and training strategies. The Feature Projection Network (FPnet) presented herein is designed to learn this signal-to-image transformation through data-driven learning rather than through direct use of linear reconstruction. To further improve reconstruction results, our method integrates an image post-processing network (U-net). Experiments show that the proposed method can achieve high reconstruction quality from limited-view data with sparse measurements. When employing GPU acceleration, this method can achieve a reconstruction speed of 15 frames per second.

Metabolites from Two Dominant Thermophilic Fungal Species Thermomyces lanuginosus and Scytalidium thermophilum.

Thermomyces lanuginosus and Scytalidium thermophilum are among the most ubiquitous thermophilic fungi in compost and soil. Chemical study on these two prevalent strains collected from Yunnan led to isolation of 23 metabolites, including one new metabolite, therlanubutanolide, and 15 known compounds, isolated from the YGP culture broth of Thermomyces lanuginosus and 7 known compounds isolated from Scytalidium thermophilum, respectively. Therlanubutanolide shared the quite similar features of the same carbon skeleton and saturation as natural hexadecanoic acids. This was the first reported discovery of such a lactone as natural occurring metabolite. All the compounds were reported for the first time from thermophilic fungi. Among them, N-[(2S,3R,4E,8E)-1,3-dihydroxy-9-methyloctadeca-4,8-dien-2-yl]acetamide was for the first time reported to be a naturally occurring metabolite and its NMR data was first provided in this study. A type of PKS-derived metabolites, three 3,4-dihydronaphthalen-1(2H)-ones, which were widely found in plant pathogenic fungi as phytotoxins and reported to have antimicrobial activity, were obtained from both dominant thermophilic fungi. The frequent occurrence of such PKS phytotoxins in these two thermophilic fungi might suggest particular ecological interest.