RSRui Santos

Rui Santos

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Rui Santos has a total of 33 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Compact vision models match domain-specific foundation models for several retinal imaging classification tasks: A systematic benchmark

    by Dávid Isztl, Tahm Spitznagel, Gábor Márk Somfai, Rui Santos

    Large domain-specific foundation models have been widely adopted for retinal image analysis, yet systematic evidence for their advantage over compact general-purpose architectures remains scarce. We benchmarked nine model configurations spanning 22.8M to 303M parameters (vision transformers, hierarchical Swin Transformers, ConvNeXt, and the domain-specific RETFound models) across four tasks: OCT 8-class disease classification, and three fundus photography tasks (DME severity, glaucoma detection, and DR severity grading). All models were evaluated under identical training conditions, with both pretrained (on natural-domain image datasets) and from-scratch initializations compared using Mann-Whitney U tests. Pretraining improved accuracy by 5.18–18.41 percentage points across all tasks (p < 0.05 throughout), with larger benefits for CFP modalities and harder tasks. Compact hierarchical models (27–29M parameters) matched or exceeded larger architectures on three of four tasks. For instance, the SwinV2-tiny architecture ranked first on OCT, DME, and GL classification. The domain-specific RETFound model (303M) achieved the highest accuracy only on the most challenging task (DR severity grading, where the most severe class is underrepresented at 8% of images), where it outperformed the best compact model by 1.54 percentage points. These results indicate that compact general-purpose models may be sufficient for most retinal classification benchmarks, and that domain-specific foundation models may add higher value mainly for severity grading tasks with skewed class distributions.


  • Open Access

    Development of a cloud framework for training and deployment of deep learning models in Radiology: automatic segmentation of the human spine from CT-scans as a case-study

    by Rui Santos, Nicholas Bünger, Benedikt Herzog, Sebastiano Caprara

    Advancements in artificial intelligence (AI) and the digitalization of healthcare are revolutionizing clinical practices, with the deployment of AI models playing a crucial role in enhancing diagnostic accuracy and treatment outcomes. Our current study aims at bridging image data collected in a clinical setting, with deployment of deep learning algorithms for the segmentation of the human spine. The developed pipeline takes a decentralized approach, where selected clinical images are sent to a trusted research environment, part of private tenant in a cloud service provider. As a use-case scenario, we used the TotalSegmentator CT-scan dataset, along with its annotated ground-truth spine data, to train a ResSegNet model native to the MONAI-Label framework. Training and validation were conducted using high performance GPUs available on demand in the Trusted Research Environment. Segmentation model performance benchmarking involved metrics such as dice score, intersection over union, accuracy, precision, sensitivity, specificity, bounding F1 score, Cohen’s kappa, area under the curve, and Hausdorff distance. To further assess model robustness, we also trained a state-of-the-art nnU-Net model using the same dataset and compared both models with a pre-trained spine segmentation model available within MONAI-Label. The ResSegNet model, deployable via MONAI-Label, demonstrated performance comparable to the state-of-the-art nnU-Net framework, with both models showing strong results across multiple segmentation metrics. This study successfully trained, evaluated and deployed a decentralized deep learning model for CT-scan spine segmentation in a cloud environment. This new model was validated against state-of-the-art alternatives. This comprehensive comparison highlights the value of the MONAI-Label as an effective tool for label generation, model training, and deployment, further highlighting its user-friendly nature and ease of deployment in clinical and research settings. Further we also demonstrate that such tools can be deployed in private and safe decentralized cloud environments for clinical use. Author Summary In the rapidly evolving field of medical imaging, the integration of artificial intelligence (AI) and cloud computing is becoming increasingly critical for advancing diagnostic and treatment capabilities. To address the growing demand for flexible digital frameworks in the clinical environment supporting the deployment of data-driven applications, we have developed and deployed a cloud-based Trusted Research Environment designed specifically for training, validation, and deployment of deep learning models focused on semantic segmentation in musculoskeletal radiology. This environment facilitates the efficient handling of large datasets and the accessibility of algorithmic output for the physicians, optimizing the interface between development and clinical translation. The established framework enables significant improvements in the deployment of deep learning tools for image analysis in the clinical setting. In our current use-case, we have utilized this environment to train and evaluate two advanced deep learning models for the segmentation of the human spine from CT scans. By leveraging the computational power and flexibility of the cloud-based infrastructure, we were able to perform rigorous training and comparison of these models, aiming to enhance the accuracy and reliability of spine segmentation in clinical practice. This approach not only streamlines the process of model development but also provides valuable insights into the performance and potential clinical applications of these AI-driven segmentation tools.


  • Open Access

    Combined nanometric and phylogenetic analysis of unique endocytic compartments in Giardia lamblia sheds light on the evolution of endocytosis in Metamonada

    by Rui Santos, Ásgeir Ástvaldsson, Shweta V. Pipaliya, Jon Paulin Zumthor, Joel B. Dacks, Staffan Svärd, Adrian B. Hehl, Carmen Faso

    Background: Giardia lamblia, a parasitic protist of the Metamonada supergroup, has evolved one of the most diverged endocytic compartment systems investigated so far. Peripheral endocytic compartments, currently known as peripheral vesicles or vacuoles (PVs), perform bulk uptake of fluid phase material which is then digested and sorted either to the cell cytosol or back to the extracellular space.Results: Here, we present a quantitative morphological characterization of these organelles using volumetric electron microscopy and super-resolution microscopy (SRM). We defined a morphological classification for the heterogenous population of PVs and performed a comparative analysis of PVs and endosome-like organelles in representatives of phylogenetically related taxa,Spironucleusspp. andTritrichomonas foetus. To investigate the as-yet insufficiently understood connection between PVs and clathrin assemblies inG. lamblia, we further performed an in-depth search for two key elements of the endocytic machinery, clathrin heavy chain (CHC) and clathrin light chain (CLC), across different lineages in Metamonada. Our data point to the loss of a bona fide CLC in the last Fornicata common ancestor (LFCA) with the emergence of a protein analogous to CLC (GlACLC) in theGiardiagenus. Finally, the location of clathrin in the various compartments was quantified.Conclusions: Taken together, this provides the first comprehensive nanometric view ofGiardia’s endocytic system architecture and sheds light on the evolution ofGlACLC analogues in the Fornicata supergroup and, specific to Giardia, as a possible adaptation to the formation and maintenance of stable clathrin assemblies at PVs.


  • Open Access

    Combined nanometric and phylogenetic analysis of unique endocytic compartments in Giardia lamblia sheds light on the evolution of endocytosis in Fornicata

    by Rui Santos, Ásgeir Ástvaldsson, Shweta V. Pipaliya, Jon Paulin Zumthor, Joel B. Dacks, Staffan Svärd, Adrian B. Hehl, Carmen Faso

    Giardia lamblia , a parasitic protist of the metamonada supergroup, has evolved one of the most diverged endocytic compartment systems investigated so far. Peripheral endocytic compartments, currently known as peripheral vesicles or vacuoles (PVs), perform bulk uptake of fluid phase material which is then digested and sorted either to the cell cytosol or back to the extracellular space. Here, we present a quantitative morphological characterization of these organelles using volumetric electron microscopy and super-resolution microscopy (SRM). We defined a morphological classification for the heterogenous population of PVs and performed a comparative analysis of PVs and endosome-like organelles in representatives of phylogenetically-related taxa, Spironucleus spp. and Tritrichomonas foetus . To investigate the as-yet insufficiently understood connection between PVs and clathrin assemblies in G. lamblia , we further performed an in-depth search for two key elements of the endocytic machinery, clathrin heavy chain (CHC) and clathrin light chain (CLC) across different lineages in Metamonada. Our data point to the loss of a bona fide CLC in the last Fornicata common ancestor (LFCA) with the emergence of a protein analogous to CLC ( Gl ACLC) in the Giardia genus. Taken together, this provides the first comprehensive nanometric view of Giardia ’s endocytic system architecture and sheds light on the evolution of GLACLC analogues in the Fornicata supergroup and, specific to Giardia, as a possible adaptation to the formation and maintenance of stable clathrin assemblies at PVs.


  • Open Access

    Unexpected organellar locations of ESCRT machinery in Giardia intestinalis and complex evolutionary dynamics spanning the transition to parasitism in the lineage Fornicata

    by Shweta V. Pipaliya, Rui Santos, Dayana Salas-Leiva, Erina A. Balmer, Corina D. Wirdnam, Andrew J. Roger, Adrian B. Hehl, Carmen Faso, Joel B. Dacks

    Background: Comparing a parasitic lineage to its free-living relatives is a powerful way to understand how that evolutionary transition to parasitism occurred. Giardia intestinalis (Fornicata) is a leading cause of gastrointestinal disease world-wide and is famous for its unusual complement of cellular compartments, such as having peripheral vacuoles instead of typical endosomal compartments. Endocytosis plays an important role in Giardia’s pathogenesis. Endosomal sorting complexes required for transport (ESCRT) are membrane-deforming proteins associated with the late endosome/multivesicular body (MVB). MVBs are ill-defined in G. intestinalis, and roles for identified ESCRT-related proteins are not fully understood in the context of its unique endocytic system. Furthermore, components thought to be required for full ESCRT functionality have not yet been documented in this species. Results: We used genomic and transcriptomic data from several Fornicata species to clarify the evolutionary genome streamlining observed in Giardia, as well as to detect any divergent orthologs of the Fornicata ESCRT subunits. We observed differences in the ESCRT machinery complement between Giardia strains. Microscopy-based investigations of key components of ESCRT machinery such as GiVPS36 and GiVPS25 link them to peripheral vacuoles, highlighting these organelles as simplified MVB equivalents. Unexpectedly, we show ESCRT components associated with the endoplasmic reticulum and, for the first time, mitosomes. Finally, we identified the rare ESCRT component CHMP7 in several fornicate representatives, including Giardia and show that contrary to current understanding, CHMP7 evolved from a gene fusion of VPS25 and SNF7 domains, prior to the last eukaryotic common ancestor, over 1.5 billion years ago. Conclusions: Our findings show that ESCRT machinery in G. intestinalis is far more varied and complete than previously thought, associates to multiple cellular locations, and presents changes in ESCRT complement which pre-date adoption of a parasitic lifestyle.


  • Open Access

    Unexpected organellar locations of ESCRT machinery in Giardia intestinalis and complex evolutionary dynamics spanning the transition to parasitism in the lineage Fornicata

    by Shweta V. Pipaliya, Rui Santos, Dayana Salas-Leiva, Erina A. Balmer, Corina D. Wirdnam, Andrew J. Roger, Adrian B. Hehl, Carmen Faso, Joel B. Dacks

    Comparing a parasitic lineage to its free-living relatives is a powerful way to understand how the evolutionary transition to parasitism occurred. Giardia intestinalis (Fornicata) is a leading cause of gastrointestinal disease world-wide and is famous for its unusual complement of cellular compartments, such as having peripheral vacuoles instead of typical endosomal compartments. Endocytosis plays an important role in Giardia ’s pathogenesis. Endosomal sorting complexes required for transport (ESCRT) are membrane-deforming proteins associated with the late endosome/multivesicular body (MVB). MVBs are ill-defined in G. intestinalis and roles for identified ESCRT-related proteins are not fully understood in the context of its unique endocytic system. Furthermore, components thought to be required for full ESCRT functionality have not yet been documented in this species. We used genomic and transcriptomic data from several Fornicata species to clarify the evolutionary genome streamlining observed in Giardia , as well as to detect any divergent orthologs of the Fornicata ESCRT subunits. We observed differences in the ESCRT machinery complement between Giardia strains. Microscopy-based investigations of key components of ESCRT machinery such as Gi VPS36and Gi VPS25 link them to peripheral vacuoles, highlighting these organelles as simplified MVB equivalents. Unexpectedly, we show ESCRT components associated with the Endoplasmic Reticulum, and for the first time, mitosomes. Finally, we identified the rare ESCRT component CHMP7 in several fornicate representatives, including Giardia , and show that contrary to current understanding, CHMP7 evolved from a gene fusion of VPS25 and SNF7 domains, prior to the last eukaryotic common ancestor, over 1.5 billion years ago. Our findings show that ESCRT machinery in G. intestinalis is far more varied and complete than previously thought, and associating to multiple cellular locations and presenting changes in ESCRT complement which pre-date adoption of a parasitic lifestyle.

  • Open Access

    A New Determinant of Candida glabrata Virulence: The Acetate Exporter CgDtr1

    by Daniela Romão, Mafalda Cavalheiro, Dalila Mil-Homens, Rui Santos, Pedro Pais, Catarina Costa, Azusa Takahashi-Nakaguchi, Arsénio M. Fialho, Hiroji Chibana, Miguel C. Teixeira

    Persistence and virulence of Candida glabrata infections are multifactorial phenomena, whose understanding is crucial to design more suitable therapeutic strategies. In this study, the putative multidrug transporter CgDtr1, encoded by ORF CAGL0M06281g, is identified as a determinant of C. glabrata virulence in the infection model Galleria mellonella. CgDTR1 deletion is shown to decrease the ability to kill G. mellonella larvae by decreasing C. glabrata ability to proliferate in G. mellonella hemolymph, and to tolerate the action of hemocytes. The possible role of CgDtr1 in the resistance to several stress factors that underlie death induced by phagocytosis was assessed. CgDTR1 was found to confer resistance to oxidative and acetic acid stress. Consistently, CgDtr1 was found to be a plasma membrane acetic acid exporter, relieving the stress induced upon C. glabrata cells within hemocytes, and thus enabling increased proliferation and virulence against G. mellonella larvae.


  • The multidrug resistance transporters CgTpo1_1 and CgTpo1_2 play a role in virulence and biofilm formation in the human pathogenCandida glabrata

    by Rui Santos, Catarina Costa, Dalila Mil-Homens, Daniela Romão, Carla C.C.R. de Carvalho, Pedro Pais, Nuno P. Mira, Arsénio M. Fialho, Miguel C. Teixeira

    The mechanisms of persistence and virulence associated with Candida glabrata infections are poorly understood, limiting the ability to fight this fungal pathogen. In this study, the multidrug resistance transporters CgTpo1_1 and CgTpo1_2 are shown to play a role in C. glabrata virulence. The survival of the infection model Galleria mellonella, infected with C. glabrata, was found to increase upon the deletion of either CgTPO1_1 or CgTPO1_2. The underlying mechanisms were further explored. In the case of CgTpo1_1, this phenotype was found to be consistent with the observation that it confers resistance to antimicrobial peptides (AMP), such as the human AMP histatin‐5. The deletion of CgTPO1_2, on the other hand, was found to limit the survival of C. glabrata cells when exposed to phagocytosis and impair biofilm formation. Interestingly, CgTPO1_2 expression was found to be up‐regulated during biofilm formation, but and its deletion leads to a decreased expression of adhesin‐encoding genes during biofilm formation, which is consistent with a role in biofilm formation. CgTPO1_2 expression was further seen to decrease plasma membrane potential and affect ergosterol and fatty acid content. Altogether, CgTpo1_1 and CgTpo1_2 appear to play an important role in the virulence of C. glabrata infections, being at the cross‐road between multidrug resistance and pathogenesis.


  • Open Access

    New Mechanisms of Flucytosine Resistance in C. glabrata Unveiled by a Chemogenomics Analysis in S. cerevisiae

    by Catarina Costa, Andreia Ponte, Pedro Pais, Rui Santos, Mafalda Cavalheiro, Takashi Yaguchi, Hiroji Chibana, Miguel Cacho Teixeira

    5-Flucytosine is currently used as an antifungal drug in combination therapy, but fungal pathogens are rapidly able to develop resistance against this drug, compromising its therapeutic action. The understanding of the underlying resistance mechanisms is crucial to deal with this problem. In this work, the S. cerevisiae deletion mutant collection was screened for increased resistance to flucytosine. Through this chemogenomics analysis, 183 genes were found to confer resistance to this antifungal agent. Consistent with its known effect in DNA, RNA and protein synthesis, the most significant Gene Ontology terms over-represented in the list of 5-flucytosine resistance determinants are related to DNA repair, RNA and protein metabolism. Additional functional classes include carbohydrate and nitrogen—particularly arginine—metabolism, lipid metabolism and cell wall remodeling. Based on the results obtained for S. cerevisiae as a model system, further studies were conducted in the pathogenic yeast Candida glabrata. Arginine supplementation was found to relieve the inhibitory effect exerted by 5-flucytosine in C. glabrata. Lyticase susceptibility was found to increase within the first 30min of 5-flucytosine exposure, suggesting this antifungal drug to act as a cell wall damaging agent. Upon exponential growth resumption in the presence of 5-flucytosine, the cell wall exhibited higher resistance to lyticase, suggesting that cell wall remodeling occurs in response to 5-flucytosine. Additionally, the aquaglyceroporin encoding genes CgFPS1 and CgFPS2, from C. glabrata, were identified as determinants of 5-flucytosine resistance. CgFPS1 and CgFPS2 were found to mediate 5-flucytosine resistance, by decreasing 5-flucytosine accumulation in C. glabrata cells.