Recent successes in image analysis with deep neural networks are achieved almost exclusively with Convolutional Neural Networks (CNNs), typically trained using the backpropagation (BP) algorithm. In a 2022 preprint, Geoffrey Hinton proposed the Forward–Forward (FF) algorithm as a biologically inspired alternative, where positive and negative examples are jointly presented to the network and training is guided by a locally defined goodness function. Here, we extend the FF paradigm to CNNs. We introduce two spatially extended labeling strategies, based on Fourier patterns and morphological transformations, that enable convolutional layers to access label information across all spatial positions. On CIFAR10, we show that deeper FF-trained CNNs can be optimized successfully and that morphology-based labels prevent shortcut solutions on dataset with more complex and fine features. On CIFAR100, carefully designed label sets scale effectively to 100 classes. Class Activation Maps reveal that FF-trained CNNs learn meaningful and complementary features across layers. Together, these results demonstrate that FF training is feasible beyond fully connected networks, provide new insights into its learning dynamics and stability, and highlight its potential for neuromorphic computing and biologically inspired learning.

The possibility to shape stimulus-responsive optical polymers, especially hydrogels, by means of laser 3D printing and ablation is fostering a new concept of “smart” micro-devices that can be used for imaging, thermal stimulation, energy transducing and sensing. The composition of these polymeric blends is an essential parameter to tune their properties as actuators and/or sensing platforms and to determine the elasto-mechanical characteristics of the printed hydrogel. In light of the increasing demand for micro-devices for nanomedicine and personalized medicine, interest is growing in the combination of composite and hybrid photo-responsive materials and digital micro-/nano-manufacturing. Existing works have exploited multiphoton laser photo-polymerization to obtain fine 3D microstructures in hydrogels in an additive manufacturing approach or exploited laser ablation of preformed hydrogels to carve 3D cavities. Less often, the two approaches have been combined and active nanomaterials have been embedded in the microstructures. The aim of this review is to give a short overview of the most recent and prominent results in the field of multiphoton laser direct writing of biocompatible hydrogels that embed active nanomaterials not interfering with the writing process and endowing the biocompatible microstructures with physically or chemically activable features such as photothermal activity, chemical swelling and chemical sensing.

Tumor organoids are three-dimensional in vitro models which can recapitulate the complex mutational landscape and tissue architecture observed in cancer patients, providing a realistic model for testing novel therapies, including immunotherapies. A significant challenge in organoid research in oncology lies in developing efficient and reliable methods for segmenting organoid images, quantifying organoid growth, regression and response to treatments, as well as predicting the behavior of organoid systems. Up to now, a curated dataset of organoids co-cultured with immune cells is not available. To address this gap, we present a new public dataset, comprising both phase-contrast images of murine and patient-derived tumor organoids of one of the deadliest cancer types, the Pancreatic Ductal Adenocarcinoma, co-cultured with immune cells, and state-of-the-art algorithms for object detection and segmentation. Our dataset, OrganoIDNetData, encompassing 180 images with 33906 organoids, can be a potential common benchmark for different organoids segmentation protocols, moving beyond the current practice of training and testing these algorithms on isolated datasets.

Diagnosis of rare neuromuscular diseases often relies on muscle biopsy analysis, which varies based on the evaluator's experience. Advances in deep learning show promise in improving diagnostic accuracy by identifying standardized features and phenotypic expressions in biopsy images. Explainable artificial intelligence extracts these features from the neural network's "black box," ensuring compliance with European ethical standards for the use of clinical data in real-world applications. This study proposes a clinic-friendly workflow, based on explainable artificial intelligence. It combines forward-forward and backpropagation-trained convolutional networks to identify complementary features of Duchenne Muscular Dystrophy. Our proposal sets the forward-forward training, applied here for the first time on biomedical images, as a potential new standard for interpretable deep learning applications in clinics. We analyzed a multiphoton microscopy dataset of 1600 images from 16 human muscle biopsies obtained during elective spinal surgery, combining autofluorescence, second and third harmonic generation signals. Class Activation Maps unveiled the dual decision-making process of the convolutional network, independently trained with both standard backpropagation and forward-forward algorithms. We evaluated the significance of the discovered features by using the Mann Whitney method. Entire biopsies were analyzed by providing an attention metric, computed as the weighted mean of all significant parameters. Backpropagation, gold standard for 35 years, and forward-forward achieved over 90 % accuracy in distinguishing healthy and diseased patients tissue. Class activation maps revealed that, when trained independently with both algorithms, the same network identifies Duchenne Muscular Dystrophy tissue by focusing on different features. Both methods identified intramuscular collagen as a key feature. Backpropagation also highlighted collagen waviness and fatty tissue. Forward-forward emphasized collagen density. We integrated these complementary insights, validated by significance analysis, into a standardized attention metric, allowing a multi-structural, quantitative assessment of tissue changes and highlighting areas for further clinical analysis. Our workflow, using clinically routine biopsies and transparent diagnostic features, demonstrates the unique potential of forward-forward in providing novel, reliable, interpretable results from biomedical images. This paves the way for its dual use with backpropagation, shifting benchmarks by enabling the discovery of features potentially overlooked by backpropagation across biomedical datasets.

Tumor organoids are three-dimensional in vitro models which can recapitulate the complex mutational landscape and tissue architecture observed in cancer patients, providing a realistic tumor microenvironment for testing novel therapies, including immunotherapies. A significant challenge in organoid research in oncology lies in developing efficient and reliable methods for segmenting organoid images, quantifying organoid growth, regression and response to treatments, as well as predicting the behavior of organoid systems. Up to now, a curated dataset of organoids co-cultured with immune cells is not available. To address this gap, we present a new public dataset, comprising both phase-contrast images of murine and patient-derived tumor organoids of one of the deadliest cancer types, the Pancreatic Ductal Adenocarcinoma, co-cultured with immune cells, and state-of-the-art algorithms for object detection and segmentation. Our dataset, OrganoIDNetData , encompassing 190 images with 33906 organoids, can be a potential common benchmark for different organoids segmentation protocols, moving beyond the current practice of training and testing these algorithms on isolated datasets.

Under the Paris Agreement, countries are encouraged to preserve and enhance existing carbon sinks, especially forests, thereby including the LULUCF (Land Use, Land Use Change and Forestry) sector in international climate mitigation targets. In particular, Europe has set the target to reach climate neutrality, i.e. a balance between anthropogenic emissions by sources and removals by sinks, by 2050. A prerequisite to reach these goals is an accurate and credible estimation of both these large fluxes. However, recent works highlighted the uncertainty related to the quantification of the land sector mitigation potential, one of the most challenging emission sectors. Moreover, the definition of climate mitigation policies often occur at the local level, where details on CO2 removals from forests and other land uses are traditionally lacking. Indeed, local authorities (e.g. cities and Regions) can be more effective in the transition to a sustainable economy compared to higher level authorities such as Nations. In this study, we tested a data-driven method based on eddy covariance (EC) data to quantify the current LULUCF role to the regional carbon sink of the Aosta Valley Region (Italy), by the integration of different approaches. Our model is based on eddy covariance measurements of CO2 fluxes, MODIS NDVI (250m), daily gridded meteorological variables at 100m spatial resolution, and a land cover map at 250m spatial resolution. A Random Forest model was used to up-scale the point eddy covariance data to the Regional level, by testing different sets of drivers (air temperature, VPD, Snow (presence/absence), NDVI, solar radiation,...). Our model was then compared to independent data derived from the National Forest Inventory (NFI), and a process-based model. Preliminary results show that forests and other ecosystems of the Region remove nearly 70% of the total anthropogenic emissions in this area. The discrepancies between the different methods will be discussed by exploring the different advantages and flaws and the spatio-temporal variability of the different approaches. Such an assessment of the local carbon budget and its uncertainties will provide a solid base for Climate-smart management of the territory and thus for reaching the carbon neutrality targets.