Conductive hydrogels are emerging as promising materials for electronic implants owing to their favorable mechanical and electrical properties. Poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) hydrogels are particularly attractive, but their preparation often requires toxic additives. Here, we introduced a nutritive sweetener, d -sorbitol, as a nontoxic additive to create soft and stretchable PEDOT:PSS conductive hydrogels. These hydrogels exhibit mechanical properties comparable with biological tissues, reducing adverse immune responses. The hydrogels can be patterned on elastic substrates using a simple, low-cost micromolding technique to fabricate soft and stretchable implantable devices for electrical stimulation and recording. The hydrogel electrodes show much lower electrochemical impedance and higher charge storage and injection capacity compared to platinum electrodes. In addition, the properties of hydrogels and devices remain stable after long-term storage and exposure to extreme conditions. We demonstrate the use of soft hydrogel-based electronic devices for effective electrical stimulation and high-quality electrical recordings in live animal models.

Glioblastoma is an aggressive central nervous system tumor characterized by diffuse infiltration. Despite substantial advances in oncology, survival outcomes have shown little improvement over the past three decades. Radiotherapy remains a cornerstone of treatment; however, it faces several challenges, including considerable inter-observer variability in clinical target volume delineation, dose constraints associated with adjacent organs at risk, and the persistently poor prognosis of affected patients. Recent advances in artificial intelligence, particularly deep learning, have shown promise in automating radiation therapy mapping to improve consistency, accuracy, and efficiency. This narrative review explores current auto segmentation frameworks, dose mapping, and biologically informed radiotherapy planning guided by multimodal imaging and mathematical modeling. Studies have demonstrated reproducible tumor segmentations with DSCs exceeding 0.90, reduced planning within minutes, and emerging predictive capabilities for treatment response. Radiogenomic integration has enabled imaging-based classification of critical biomarkers with high accuracy, reinforcing the potential of deep learning models in personalized radiotherapy. Despite these innovations, deployment into clinical practice remains limited, primarily due to insufficient external validation and single-institution training datasets. This review emphasizes the importance of large, annotated imaging datasets, multi-institutional collaboration, and biologically explainable modeling to successfully translate deep learning into glioblastoma radiation planning and longitudinal monitoring.

Glioblastoma is an aggressive central nervous system tumor characterized by diffuse infiltration. Despite substantial advances in oncology, survival outcomes have shown little improvement over the past three decades. Radiotherapy remains a cornerstone of treatment; however, it faces several challenges, including considerable inter-observer variability in clinical target volume delineation, dose constraints associated with adjacent organs at risk, and the persistently poor prognosis of affected patients. Recent advances in artificial intelligence, particularly deep learning, have shown promise in automating radiation therapy mapping to improve consistency, accuracy, and efficiency. This narrative review explores current auto segmentation frameworks, dose mapping, and biologically informed radiotherapy planning guided by multimodal imaging and mathematical modeling. Studies have demonstrated reproducible tumor segmentations with DSCs exceeding 0.90, reduced planning within minutes, and emerging predictive capabilities for treatment response. Radiogenomic integration has enabled imaging-based classification of critical biomarkers with high accuracy, reinforcing the potential of deep learning models in personalized radiotherapy. Despite these innovations, deployment into clinical practice remains limited, primarily due to insufficient external validation and single-institution training datasets. This review emphasizes the importance of large, annotated imaging datasets, multi-institutional collaboration, and biologically explainable modeling to successfully translate deep learning into glioblastoma radiation planning and longitudinal monitoring.

SignificanceThyroid cancer is the most common endocrine malignancy. Although fine needle aspiration (FNA) is the gold standard for diagnosis, 10% to 20% of FNAs yield indeterminate cytology, of which 20% to 30% prove malignant. Such diagnostic uncertainty drives repeat FNAs, costly molecular tests, and unnecessary surgeries, leading to increased healthcare costs, reduced quality of life, and emotional distress to patients. Current methods provide limited structural information about the extra-cellular matrix, despite collagen remodeling being a hallmark of malignancy. A rapid, label-free, and architecture-based diagnostic approach has the potential to significantly improve real-time or near-real-time decision-making and reduce unnecessary biopsies and interventions.AimWe use a systematic, quantitative framework based on second harmonic generation (SHG) microscopy to differentiate between normal and cancerous (papillary thyroid carcinoma [PTC]) human thyroid tissues through collagen architecture signatures.ApproachFormalin-fixed paraffin-embedded unstained tissue sections (5 μm thick) of confirmed PTC and normal thyroid tissue were obtained under an approved IRB and imaged under a multiphoton microscope using an excitation wavelength of 803 nm. An emission filter of 400±10 nm was used to filter the SHG emission from the sample. Thirty-five metrics capturing intensity, geometric, textural, and frequency-domain collagen features were quantified. A linear mixed-effects model was used for quantitative comparison between cancer and normal groups, accounting for the hierarchical structure of the data and rigorously assessing metric significance despite limited sample size.ResultsOf the 35 metrics evaluated, 19 remained statistically significant after false discovery rate (FDR) adjustment for multiple comparisons. The corresponding effect sizes ranged from moderate to large, indicating substantial separation between normal and cancerous samples. Five of the remaining 16 metrics, although not statistically significant, were consistently either higher or lower in one group compared with the other, suggesting directional consistency. This revealed robust and biologically coherent alterations in collagen organization despite small cohort size. These included consistent increases in spectral power across multiple spatial frequency bands and greater collagen heterogeneity in cancer tissue.ConclusionsThe convergence of significant and directionally consistent metrics underscores the strong diagnostic potential of SHG-derived collagen architecture. Integrating these quantitative features into an automated analysis platform could enable rapid, objective, and reproducible optical biopsies, with the potential to reduce indeterminate FNAs, guide intraoperative decision-making, and improve diagnostic accuracy in routine thyroid cancer care.