Visualising and quantifying relevant parkinsonian gait patterns using 3D convolutional network.

Parkinson's disease (PD) lacks a definitive diagnosis, with the observation of motion patterns being the main method of characterizing disease progression and planning patient treatments. Among PD observations, gait motion patterns, such as step length, flexed posture, and bradykinesia, support the characterization of disease progression. However, this analysis is usually performed with marker-based protocols, which affect the gait and localized segment patterns during locomotion. This work introduces a 3D convolutional gait representation for automatic PD classification that identifies the spatio-temporal patterns used for classification. This approach allows us to obtain an explainable model that classifies markerless sequences and describes the main learned spatio-temporal regions associated with abnormal patterns in a particular video. Initially, a spatio-temporal convolutional network is trained from a set of raw videos and optical flow fields. Then, a PD prediction is obtained from the motion patterns learned by the trained model. Finally, saliency maps, which highlight abnormal motion patterns, are obtained by retro-propagating the output prediction up to the input volume through two different stages: an embedded back-tracking and a pseudo-deconvolution process. From a total of 176 videos from 22 patients, the resulting salient maps highlight lower limb patterns possibly related to step length and speed. In control subjects, the saliency maps highlight the head and trunk posture. The proposed approach achieved an average accuracy score of 94.89%.

Analysis of local genome rearrangement improves resolution of ancestral genomic maps in plants.

BACKGROUND: Computationally inferred ancestral genomes play an important role in many areas of genome research. We present an improved workflow for the reconstruction from highly diverged genomes such as those of plants. RESULTS: Our work relies on an established workflow in the reconstruction of ancestral plants, but improves several steps of this process. Instead of using gene annotations for inferring the genome content of the ancestral sequence, we identify genomic markers through a process called genome segmentation. This enables us to reconstruct the ancestral genome from hundreds of thousands of markers rather than the tens of thousands of annotated genes. We also introduce the concept of local genome rearrangement, through which we refine syntenic blocks before they are used in the reconstruction of contiguous ancestral regions. With the enhanced workflow at hand, we reconstruct the ancestral genome of eudicots, a major sub-clade of flowering plants, using whole genome sequences of five modern plants. CONCLUSIONS: Our reconstructed genome is highly detailed, yet its layout agrees well with that reported in Badouin et al. (2017). Using local genome rearrangement, not only the marker-based, but also the gene-based reconstruction of the eudicot ancestor exhibited increased genome content, evidencing the power of this novel concept.

Computing the family-free DCJ similarity.

BACKGROUND: The genomic similarity is a large-scale measure for comparing two given genomes. In this work we study the (NP-hard) problem of computing the genomic similarity under the DCJ model in a setting that does not assume that the genes of the compared genomes are grouped into gene families. This problem is called family-free DCJ similarity. RESULTS: We propose an exact ILP algorithm to solve the family-free DCJ similarity problem, then we show its APX-hardness and present four combinatorial heuristics with computational experiments comparing their results to the ILP. CONCLUSIONS: We show that the family-free DCJ similarity can be computed in reasonable time, although for larger genomes it is necessary to resort to heuristics. This provides a basis for further studies on the applicability and model refinement of family-free whole genome similarity measures.

Design, development and validation of a new laryngo-pharyngeal endoscopic esthesiometer and range-finder based on the assessment of air-pulse variability determinants.

BACKGROUND: Laryngo-pharyngeal mechano-sensitivity (LPMS) is involved in dysphagia, sleep apnea, stroke, irritable larynx syndrome and cough hypersensitivity syndrome among other disorders. These conditions are associated with a wide range of airway reflex abnormalities. However, the current device for exploring LPMS is limited because it assesses only the laryngeal adductor reflex during fiber-optic endoscopic evaluations of swallowing and requires a high degree of expertise to obtain reliable results, introducing intrinsic expert variability and subjectivity. METHODS: We designed, developed and validated a new air-pulse laryngo-pharyngeal endoscopic esthesiometer with a built-in laser range-finder (LPEER) based on the evaluation and control of air-pulse variability determinants and on intrinsic observer variability and subjectivity determinants of the distance, angle and site of stimulus impact. The LPEER was designed to be capable of delivering precise and accurate stimuli with a wide range of intensities that can explore most laryngo-pharyngeal reflexes. RESULTS: We initially explored the potential factors affecting the reliability of LPMS tests and included these factors in a multiple linear regression model. The following factors significantly affected the precision and accuracy of the test (P < 0.001): the tube conducting the air-pulses, the supply pressure of the system, the duration of the air-pulses, and the distance and angle between the end of the tube conducting the air-pulses and the site of impact. To control all of these factors, an LPEER consisting of an air-pulse generator and an endoscopic laser range-finder was designed and manufactured. We assessed the precision and accuracy of the LPEER's stimulus and range-finder according to the coefficient of variation (CV) and by looking at the differences between the measured properties and the desired values, and we performed a pilot validation on ten human subjects. The air-pulses and range-finder exhibited good precision and accuracy (CV < 0.06), with differences between the desired and measured properties at <3 % and a range-finder measurement error of <1 mm. The tests in patients demonstrated obtainable and reproducible thresholds for the laryngeal adductor, cough and gag reflexes. CONCLUSIONS: The new LPEER was capable of delivering precise and accurate stimuli for exploring laryngo-pharyngeal reflexes.

On the distribution of cycles and paths in multichromosomal breakpoint graphs and the expected value of rearrangement distance.

Finding the smallest sequence of operations to transform one genome into another is an important problem in comparative genomics. The breakpoint graph is a discrete structure that has proven to be effective in solving distance problems, and the number of cycles in a cycle decomposition of this graph is one of the remarkable parameters to help in the solution of related problems. For a fixed k, the number of linear unichromosomal genomes (signed or unsigned) with n elements such that the induced breakpoint graphs have k disjoint cycles, known as the Hultman number, has been already determined. In this work we extend these results to multichromosomal genomes, providing formulas to compute the number of multichromosal genomes having a fixed number of cycles and/or paths. We obtain an explicit formula for circular multichromosomal genomes and recurrences for general multichromosomal genomes, and discuss how these series can be used to calculate the distribution and expected value of the rearrangement distance between random genomes.

Production of bacteriocin-like inhibitory substance by Bifidobacterium lactis in skim milk supplemented with additives.

Bacteriocins are natural compounds used as food biopreservatives instead of chemical preservatives. Bifidobacterium animalis subsp. lactis (Bifid. lactis) was shown to produce a bacteriocin-like inhibitory substance (BLIS) able to inhibit the growth of Listeria monocytogenes selected as an indicator microorganism. To enhance this production by the strain Bifid. lactis BL 04, skim milk (SM) was used as a fermentation medium either in the presence or in the absence of yeast extract, Tween 80 or inulin as stimulating additives, and the results in terms of bacterial growth and BLIS production were compared with those obtained in a traditional high cost complex medium such as Man, Rogosa and Sharpe (MRS). To this purpose, all the cultivations were carried out in flasks at 200 rpm under anaerobic conditions ensured by a nitrogen flowrate of 1.0 L/min for 48 h, and BLIS production was quantified by means of a modified agar diffusion assay at low values of both temperature and concentration of List. monocytogenes. Although all these ingredients were shown to exert positive influence on BLIS production in both media, yeast extract and SM were by far the best ingredient and the best medium, respectively, allowing for a BLIS production at the late exponential phase of 2000 AU/ml.

A non-aligned translation with a neoplastic classifier regularization to include vascular NBI patterns in standard colonoscopies.

Polyp vascular patterns are key to categorizing colorectal cancer malignancy. These patterns are typically observed in situ from specialized narrow-band images (NBI). Nonetheless, such vascular characterization is lost from standard colonoscopies (the primary attention mechanism). Besides, even for NBI observations, the categorization remains biased for expert observations, reporting errors in classification from 59.5% to 84.2%. This work introduces an end-to-end computational strategy to enhance in situ standard colonoscopy observations, including vascular patterns typically observed from NBI mechanisms. These retrieved synthetic images are achieved by adjusting a deep representation under a non-aligned translation task from optical colonoscopy (OC) to NBI. The introduced scheme includes an architecture to discriminate enhanced neoplastic patterns achieving a remarkable separation into the embedding representation. The proposed approach was validated in a public dataset with a total of 76 sequences, including standard optical sequences and the respective NBI observations. The enhanced optical sequences were automatically classified among adenomas and hyperplastic samples achieving an F1-score of 0.86%. To measure the sensibility capability of the proposed approach, serrated samples were projected to the trained architecture. In this experiment, statistical differences from three classes with a ρ-value <0.05 were reported, following a Mann-Whitney U test. This work showed remarkable polyp discrimination results in enhancing OC sequences regarding typical NBI patterns. This method also learns polyp class distributions under the unpaired criteria (close to real practice), with the capability to separate serrated samples from adenomas and hyperplastic ones.

Cardiac disease discrimination from 3D-convolutional kinematic patterns on cine-MRI sequences. / Discriminación de enfermedades cardiacas utilizando patrones cinemáticos codificados con convoluciones 3D en secuencias de cine-RM

INTRODUCTION: Cine-MRI (cine-magnetic resonance imaging) sequences are a key diagnostic tool to visualize anatomical information, allowing experts to localize and determine suspicious pathologies. Nonetheless, such analysis remains subjective and prone to diagnosis errors. OBJECTIVE: To develop a binary and multi-class classification considering various cardiac conditions using a spatiotemporal model that highlights kinematic movements to characterize each disease. MATERIALS AND METHODS: This research focuses on a 3D convolutional representation to characterize cardiac kinematic patterns during the cardiac cycle, which may be associated with pathologies. The kinematic maps are obtained from the apparent velocity maps computed from a dense optical flow strategy. Then, a 3D convolutional scheme learns to differentiate pathologies from kinematic maps. RESULTS: The proposed strategy was validated with respect to the capability to discriminate among myocardial infarction, dilated cardiomyopathy, hypertrophic cardiomyopathy, abnormal right ventricle, and normal cardiac sequences. The proposed method achieves an average accuracy of 78.00% and a F1 score of 75.55%. Likewise, the approach achieved 92.31% accuracy for binary classification between pathologies and control cases. CONCLUSION: The proposed method can support the identification of kinematically abnormal patterns associated with a pathological condition. The resultant descriptor, learned from the 3D convolutional net, preserves detailed spatiotemporal correlations and could emerge as possible digital biomarkers of cardiac diseases.

Introducción. Las secuencias del cine-resonancia magnética (cine-MRI, cine magnetic resonance imaging) son una herramienta diagnóstica clave para visualizar la información anatómica que les permite a los expertos localizar y determinar aquellas anomalías que resulten sospechosas. No obstante, este análisis sigue siendo subjetivo y propenso a errores de diagnóstico. Objetivo. Desarrollar una clasificación binaria y multiclase, considerando diferentes condiciones cardiacas, mediante un modelo espaciotemporal que permita resaltar los movimientos cinéticos para caracterizar cada enfermedad. Materiales y métodos. Este estudio se centra en el uso de una representación de convolución 3D para caracterizar los patrones cinéticos durante el ciclo cardiaco que puedan estar asociados con enfermedades. Para ello, se obtienen mapas cinéticos a partir de mapas de velocidad aparente, calculados mediante una estrategia de flujo óptico denso. A continuación, un esquema de convolución 3D "aprende" a diferenciar patologías a partir de mapas cinemáticos. Resultados. La estrategia propuesta se validó según la capacidad de discriminar entre infarto de miocardio, miocardiopatía dilatada, miocardiopatía hipertrófica, ventrículo derecho anormal y un examen normal. El método propuesto alcanza una precisión media del 78,0 % y una puntuación F1 score del 75,55 %. Asimismo, el enfoque alcanzó el 92,31 % de precisión para la clasificación binaria entre enfermedades y casos de control. Conclusiones. El método propuesto es capaz de apoyar la identificación de patrones cinéticos anormales asociados con una condición patológica. El descriptor resultante, aprendido de la red de convolución 3D, conserva correlaciones espaciotemporales detalladas y podría surgir como posible biomarcador digital de enfermedades cardiacas.

Dual-process model of courage.

Courage is one of the most significant psychological constructs for society, but not one of the most frequently studied. This paper presents a process model of courage consisting of decision-based pathways by which one comes to enact a courageous action. We argue the process of courage begins with a trigger involving an actor(s) and a situation(s). The actor(s) then engage(s) in four key assessments concerning (a) immediacy of the situation, (b) meaningfulness, value, and relevance to the actor, (c) adequacy of efficacy to act, and (d) decision to act with courage. The central component of this process entails an approach-avoidance conflict involving assessments of perceived risks and potential noble outcomes of acting with courage. The decision to act may result in courageous actions assuming it satisfies the four elements: intentionality, objective and substantial risk, a noble purpose, and meaning in time and place. Courageous actions have consequences. Finally, the consequences shape the actors' experience, which feeds into the trigger, closing the loop. Potential moderators of the courage process as well as potential tests of the model have been discussed.

APIS: a paired CT-MRI dataset for ischemic stroke segmentation - methods and challenges.

Stroke, the second leading cause of mortality globally, predominantly results from ischemic conditions. Immediate attention and diagnosis, related to the characterization of brain lesions, play a crucial role in patient prognosis. Standard stroke protocols include an initial evaluation from a non-contrast CT to discriminate between hemorrhage and ischemia. However, non-contrast CTs lack sensitivity in detecting subtle ischemic changes in this phase. Alternatively, diffusion-weighted MRI studies provide enhanced capabilities, yet are constrained by limited availability and higher costs. Hence, we idealize new approaches that integrate ADC stroke lesion findings into CT, to enhance the analysis and accelerate stroke patient management. This study details a public challenge where scientists applied top computational strategies to delineate stroke lesions on CT scans, utilizing paired ADC information. Also, it constitutes the first effort to build a paired dataset with NCCT and ADC studies of acute ischemic stroke patients. Submitted algorithms were validated with respect to the references of two expert radiologists. The best achieved Dice score was 0.2 over a test study with 36 patient studies. Despite all the teams employing specialized deep learning tools, results reveal limitations of computational approaches to support the segmentation of small lesions with heterogeneous density.

Simulation of normal and pathological gaits using a fusion knowledge strategy.

: Gait distortion is the first clinical manifestation of many pathological disorders. Traditionally, the gait laboratory has been the only available tool for supporting both diagnosis and prognosis, but under the limitation that any clinical interpretation depends completely on the physician expertise. This work presents a novel human gait model which fusions two important gait information sources: an estimated Center of Gravity (CoG) trajectory and learned heel paths, by that means allowing to reproduce kinematic normal and pathological patterns. The CoG trajectory is approximated with a physical compass pendulum representation that has been extended by introducing energy accumulator elements between the pendulum ends, thereby emulating the role of the leg joints and obtaining a complete global gait description. Likewise, learned heel paths captured from actual data are learned to improve the performance of the physical model, while the most relevant joint trajectories are estimated using a classical inverse kinematic rule. The model is compared with standard gait patterns, obtaining a correlation coefficient of 0.96. Additionally,themodel simulates neuromuscular diseases like Parkinson (phase 2, 3 and 4) and clinical signs like the Crouch gait, case in which the averaged correlation coefficient is 0.92.

Talent identification research: a bibliometric study from multidisciplinary and global perspectives.

This paper describes the general status, trends, and evolution of research on talent identification across multiple fields globally over the last 80 years. Using Scopus and Web of Science databases, we explored patterns of productivity, collaboration, and knowledge structures in talent identification (TI) research. Bibliometric analysis of 2,502 documents revealed talent identification research is concentrated in the fields of management, business, and leadership (~37%), sports and sports science (~20%), and education, psychology, and STEM (~23%). Whereas research in management and sports science has occurred independently, research in psychology and education has created a bridge for the pollination of ideas across fields. Thematic evolution analysis indicates that TI has well developed motor and basic research themes focused on assessment, cognitive abilities, fitness, and youth characteristics. Motor themes in management and sports science bring attention to talent management beyond TI. Emerging research focuses on equity and diversity as well as innovation in identification and technology-based selection methods. Our paper contributes to the development of the body of TI research by (a) highlighting the role of TI across multiple disciplines, (b) determining the most impactful sources and authors in TI research, and (c) tracing the evolution of TI research which identifies gaps and future opportunities for exploring and developing TI research and its broader implications for other areas of research and society.

A deep supervised cross-attention strategy for ischemic stroke segmentation in MRI studies.

The key component of stroke diagnosis is the localization and delineation of brain lesions, especially from MRI studies. Nonetheless, this manual delineation is time-consuming and biased by expert opinion. The main purpose of this study is to introduce an autoencoder architecture that effectively integrates cross-attention mechanisms, together with hierarchical deep supervision to delineate lesions under scenarios of remarked unbalance tissue classes, challenging geometry of the shape, and a variable textural representation. This work introduces a cross-attention deep autoencoder that focuses on the lesion shape through a set of convolutional saliency maps, forcing skip connections to preserve the morphology of affected tissue. Moreover, a deep supervision training scheme was herein adapted to induce the learning of hierarchical lesion details. Besides, a special weighted loss function remarks lesion tissue, alleviating the negative impact of class imbalance. The proposed approach was validated on the public ISLES2017 dataset outperforming state-of-the-art results, achieving a dice score of 0.36 and a precision of 0.42. Deeply supervised cross-attention autoencoders, trained to pay more attention to lesion tissue, are better at estimating ischemic lesions in MRI studies. The best architectural configuration was achieved by integrating ADC, TTP and Tmax sequences. The contribution of deeply supervised cross-attention autoencoders allows better support the discrimination between healthy and lesion regions, which in consequence results in favorable prognosis and follow-up of patients.

The Evolution of Intelligence: Analysis of the Journal of Intelligence and Intelligence.

What are the current trends in intelligence research? This parallel bibliometric analysis covers the two premier journals in the field: Intelligence and the Journal of Intelligence (JOI) between 2013 and 2022. Using Scopus data, this paper extends prior bibliometric articles reporting the evolution of the journal Intelligence from 1977 up to 2018. It includes JOI from its inception, along with Intelligence to the present. Although the journal Intelligence's growth has declined over time, it remains a stronghold for traditional influential research (average publications per year = 71.2, average citations per article = 17.07, average citations per year = 2.68). JOI shows a steady growth pattern in the number of publications and citations (average publications per year = 33.2, average citations per article = 6.48, total average citations per year = 1.48) since its inception in 2013. Common areas of study across both journals include cognitive ability, fluid intelligence, psychometrics-statistics, g-factor, and working memory. Intelligence includes core themes like the Flynn effect, individual differences, and geographic IQ variability. JOI addresses themes such as creativity, personality, and emotional intelligence. We discuss research trends, co-citation networks, thematic maps, and their implications for the future of the two journals and the evolution and future of the scientific study of intelligence.

A multitask deep representation for Gleason score classification to support grade annotations.

The Gleason grade system is the main standard to quantify the aggressiveness and progression of prostate cancer. Currently, exists a high disagreement among experts in the diagnosis and stratification of this disease. Deep learning models have emerged as an alternative to classify and support experts automatically. However, these models are limited to learn a rigid stratification rule that can be biased during training to a specific observer. Therefore, this work introduces an embedding representation that integrates an auxiliary task learning to deal with the high inter and intra appearance of the Gleason system. The proposed strategy implements as a main task a triplet loss scheme that builds a feature embedding space with respect to batches of positive and negative histological training patches. As an auxiliary task is added a cross-entropy that helps with inter-class variability of samples while adding robust representations to the main task. The proposed approach shows promising results achieving an average accuracy of 66% and 64%, for two experts without statistical difference. Additionally, reach and average accuracy of 73% in patches where both pathologists are agree, showing the robustness patterns learning from the approach.

Weakly Supervised Polyp Segmentation from an Attention Receptive Field Mechanism.

Colorectal cancer is the third most incidence cancer world-around. Colonoscopies are the most effective resource to detect and segment abnormal polyp masses, considered as the main biomarker of this cancer. Nonetheless, some recent clinical studies have revealed a polyp miss rate up to 26% during the clinical routine. Also, the expert bias introduced during polyp shape characterization may induce to false-negative diagnosis. Current computational approaches have supported polyp segmentation but over controlled scenarios, where polyp frames have been labeled by an expert. These supervised representations are fully dependent of well-segmented polyps, in crop sequences that always report these masses. This work introduces an attention receptive field mechanism, that robustly recover the polyp shape, by learning non-local pixel relationship. Besides this deep representation is learning from a weakly supervised scheme that includes unlabeled background frames, to discriminate polyps from near structures like intestinal folds. The achieved results outperform state-of-the-art approaches achieving a 95.1% precision in the public CVC-Colon DB, with also competitive performance on other datasets. Clinical relevance-The work address a novel strategy to support segmentation tools in a clinical routine with redundant background over colonoscopy sequences.

A Riemannian Deep Learning Representation to Describe Gait Parkinsonian Locomotor Patterns.

Parkinson is the second most common neurodegenerative disease, mainly related to progressive locomotor alterations caused by dopamine deficiency. The gait kinematic is a principal disease biomarker that associates patterns like the step length, flexed posture, and bradykinesia with disease progression. Nonetheless, these patterns are analyzed from invasive setups that only capture coarse dynamics at advanced disease stages. This work introduces a very compact and robust deep representation that effectively learns a Riemannian manifold to describe locomotor parkinsonian patterns. Contrary to traditional deep strategies, the presented framework fully explores data geometry from input mean covariance matrices representing video sequences. These symmetric positive definite (SPD) matrices lie in a Riemannian manifold. Then such matrices are projected to the SPD net, which learns a bank of SPD matrices from a non-linear training, that may be exploited in a hierarchical composition from a set of layers. In a final layer, a projection in a Euclidean space allows learning the discriminatory patterns of PD w.r.t control population. In a retrospective study with a total of 22 patients (11 Parkinson's and 11 controls), the proposed approach achieves a remarkable classification between Parkinson's and control video sequences, correctly labeling all Parkinson's patients, and outperforming typical 3D convolutional representations. Clinical relevance- The study of gait parkinsonian descrip-tors computed from a Riemannian geometry, built from a deep representation.

A Parkinsonian Digital Biomarker Learned as an Anomaly Deep Generative Representation.

Parkinson's Disease (PD), the second most common neurodegenerative disorder, is associated with voluntary movement disorders caused by progressive dopamine deficiency. Gait motor alterations constitute a main tool to diagnose, characterize and personalize treatments. Nonetheless, such evaluation is biased by expert observations, reporting a false positive diagnosis up to 24%. Learning computational tools are recently emerged as potential alternatives to support diagnosis and to quantify kinematic patterns during locomotion. Nonetheless, such learning schemes required a large amount of balanced and stratified data examples, which may result unrealistic in clinical scenarios. This work introduces a self-supervised generative representation to discover gait-motion related patterns, under the pretext of video reconstruction and an anomaly detection framework. From the learned scheme, it is recovered a hidden embedding gait descriptor that constitutes a digital biomarker, allowing to discover PD differences regarding a control population. The proposed approach was validated with 11 PD patients (H&Y scale between 2.5 and 3.0) and 11 control subjects, and trained with only control population, achieving an AUC of 99.4% in the classification task. Clinical Relevance- A digital biomarker that helps in the diagnosis of PD using videos of a patient's gait to capture important and relevant motion patterns to avoid subjectivity when an expert made a diagnosis.

A digital cardiac disease biomarker from a generative progressive cardiac cine-MRI representation.

Cardiac cine-MRI is one of the most important diagnostic tools used to assess the morphology and physiology of the heart during the cardiac cycle. Nonetheless, the analysis on cardiac cine-MRI is poorly exploited and remains highly dependent on the observer's expertise. This work introduces an imaging cardiac disease representation, coded as an embedding vector, that fully exploits hidden mapping between the latent space and a generated cine-MRI data distribution. The resultant representation is progressively learned and conditioned by a set of cardiac conditions. A generative cardiac descriptor is achieved from a progressive generative-adversarial network trained to produce MRI synthetic images, conditioned to several heart conditions. The generator model is then used to recover a digital biomarker, coded as an embedding vector, following a backpropagation scheme. Then, an UMAP strategy is applied to build a topological low dimensional embedding space that discriminates among cardiac pathologies. Evaluation of the approach is carried out by using an embedded representation as a potential disease descriptor in 2296 pathological cine-MRI slices. The proposed strategy yields an average accuracy of 0.8 to discriminate among heart conditions. Furthermore, the low dimensional space shows a remarkable grouping of cardiac classes that may suggest its potential use as a tool to support diagnosis. The learned progressive and generative representation, from cine-MRI slices, allows retrieves and coded complex descriptors that results useful to discriminate among heart conditions. The cardiac disease representation expressed as a hidden embedding vector could potentially be used to support cardiac analysis on cine-MRI sequences.

Multimodal Contrastive Supervised Learning to Classify Clinical Significance MRI Regions on Prostate Cancer.

Clinically significant regions (CSR), captured over multi-parametric MRI (mp-MRI) images, have emerged as a potential screening test for early prostate cancer detection and characterization. These sequences are able to quantify morphology, micro-circulation, and cellular density patterns that might be related to cancer disease. Nonetheless, this evaluation is mainly carried out by expert radiologists, introducing inter-reader variability in the diagnosis. Therefore, different deep learning models were proposed to support the diagnosis, but a proper representation of prostate lesions remains limited due to the non-alignment among sequences and the dependency of considerable amounts of labeled data for learning. The main limitation of such representation lies in the cross-entropy minimization that only exploits inter-class variation, being insufficient data augmentation and transfer learning strategies. This work introduces a Supervised Contrastive Learning (SCL) strategy that fully exploits the inter and intra-class variability of prostate lesions to robustly represent MRI regions. This strategy extracts lesion sample tuples, with positive and negative labels, regarding a query lesion. Such tuples are involved into an easy-positive, and semi-hard negative mining to project samples that better update the deep representation. The proposed learning strategy achieved an average ROC-AVC of 0.82, to characterize prostate cancer in MRI, using only the 60% of the available annotated data. Clinical relevance - A robust learning scheme that properly finds representations in limited data scenarios to classify clinically significant MRI regions on prostate cancer.