Automated Image Analysis Pipeline Research Articles

Combining unsupervised and supervised learning in microscopy enables defect analysis of a full 4H-SiC wafer

AbstractDetecting and analyzing various defect types in semiconductor materials is an important prerequisite for understanding the underlying mechanisms and tailoring the production processes. Analysis of microscopy images that reveal defects typically requires image analysis tasks such as segmentation and object detection. With the permanently increasing amount of data from experiments, handling these tasks manually becomes more and more impossible. In this work, we combine various image analysis and data mining techniques to create a robust and accurate, automated image analysis pipeline for extracting the type and position of all defects in a microscopy image of a KOH-etched 4H-SiC wafer. Graphical abstract

AMPK regulates phagophore-to-autophagosome maturation.

Autophagy is an important metabolic pathway that can non-selectively recycle cellular material or lead to targeted degradation of protein aggregates or damaged organelles. Autophagosome formation starts with autophagy factors accumulating on lipid vesicles containing ATG9. These phagophores attach to donor membranes, expand via ATG2-mediated lipid transfer, capture cargo, and mature into autophagosomes, ultimately fusing with lysosomes for their degradation. Autophagy can be activated by nutrient stress, for example, by a reduction in the cellular levels of amino acids. In contrast, how autophagy is regulated by low cellular ATP levels via the AMP-activated protein kinase (AMPK), an important therapeutic target, is less clear. Using live-cell imaging and an automated image analysis pipeline, we systematically dissect how nutrient starvation regulates autophagosome biogenesis. We demonstrate that glucose starvation downregulates autophagosome maturation by AMPK-mediated inhibition of phagophore tethering to donor membrane. Our results clarify AMPKs regulatory role in autophagy and highlight its potential as a therapeutic target to reduce autophagy.

VINNA for Neonates – Orientation Independence through Latent Augmentations

Abstract A robust, fast, and accurate segmentation of neonatal brain images is highly desired to better understand and detect changes during development and disease, specifically considering the rise in imaging studies for this cohort. Yet, the limited availability of ground truth datasets, lack of standardized acquisition protocols, and wide variations of head positioning in the scanner pose challenges for method development. A few automated image analysis pipelines exist for newborn brain MRI segmentation, but they often rely on time-consuming non-linear spatial registration procedures and require resampling to a common resolution, subject to loss of information due to interpolation and down-sampling. Without registration and image resampling, variations with respect to head positions and voxel resolutions have to be addressed differently. In deep-learning, external augmentations such as rotation, translation, and scaling are traditionally used to artificially expand the representation of spatial variability, which subsequently increases both the training dataset size and robustness. However, these transformations in the image space still require resampling, reducing accuracy specifically in the context of label interpolation. We recently introduced the concept of resolution-independence with the Voxel-size Independent Neural Network framework, VINN. Here, we extend this concept by additionally shifting all rigid-transforms into the network architecture with a four degree of freedom (4-DOF) transform module, enabling resolution-aware internal augmentations (VINNA) for deep learning. In this work we show that VINNA (i) significantly outperforms state-of-the-art external augmentation approaches, (ii) effectively addresses the head variations present specifically in newborn datasets, and (iii) retains high segmentation accuracy across a range of resolutions (0.5-1.0mm). Furthermore, the 4-DOF transform module together with internal augmentations is a powerful, general approach to implement spatial augmentation without requiring image or label interpolation. The specific network application to newborns will be made publicly available as VINNA4neonates.

Abstract 299: FAP/αSMA association with tumor visibility and progression in prostate cancer

Abstract Introduction Prostate cancer (PCa) ranks as the second most frequently diagnosed cancer in the male population worldwide. The challenge in effectively managing PCa is the absence of well-established biological markers, which often leads to both over and under treatment. To address this issue, a deeper understanding of the molecular alterations within PCa is essential for more accurate patient stratification. In this research, we embark on a comprehensive examination of how the presence of fibroblast activation protein (FAP) and alpha smooth muscle actin (αSMA) in PCa tissue associates with the visibility of the disease in Magnetic Resonance Imaging (MRI) and their impact on cancer progression. The objective of this study is to analyze cell compartmental expression of FAP and aSMA using an AI based image analysis method. We also study the role of FAP and αSMA expression in predicting disease progression and mortality. Materials and Methods We applied FAP (ab207178, Abcam)/αSMA (M0851, DAKO) double chromogen immunohistochemical (IHC) staining to three independent cohorts of formalin-fixed paraffin-embedded (FFPE) prostate cancer tissue microarrays. All together, we analyzed 835 patients’ samples. First cohort consists of 387 patients with preoperative multiparametric MRI and robot-assisted laparoscopic prostatectomy (RALP) as primary treatment. Second cohort was case vs control cohort of Grade group 2-4 localized PCa at radical prostatectomy (RP) with 168 patients. Third cohort consists PCa patents from continuous, population-based collection of radical prostatectomies of 319 patients. Image analysis was performed with Aiforia Create, which is cloud-based deep-learning artificial intelligence software that provides platform for training and validating automated image analysis pipelines using convolutional neural networks. Analysis algorithm in this project was based on semantic, pixel-level area segmentation. Results FAP expression was higher and αSMA lower in MRI-visible tumors compared with MRI-invisible tumors. Stromal and epithelial FAP expression associate with tumor progression. We could also see statistically significant differences in FAP positivity between clinically significant (GS ≥7) and insignificant (GS ≤6) tumors. Hazard ratios imply that FAP and αSMA have significant impact on risk of biochemical recurrence in patients with MRI-visible tumors. Conclusions FAP is closely linked to tumor MRI visibility, cancer progression, and Gleason scores, marking it as a valuable biomarker in PCa assessment. High stromal αSMA is primarily associated with MRI visibility. Our automated AI-based image analysis method has demonstrated reliability, aligning with our previous findings. This advances the potential for more accurate and efficient prostate cancer diagnosis and prognosis, offering a potential benefit to future patient care. Citation Format: Jenni Säilä, Timo-Pekka Lehto, Annabrita Schoonenberg, Katja Välimäki, Andrew Erickson, Antti Rannikko, Olli Kallioniemi, Tuomas Mirtti, Teijo Pellinen. FAP/αSMA association with tumor visibility and progression in prostate cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 299.

A high-content screen reveals new regulators of nuclear membrane stability

Nuclear membrane rupture is a physiological response to multiple in vivo processes, such as cell migration, that can cause extensive genome instability and upregulate invasive and inflammatory pathways. However, the underlying molecular mechanisms of rupture are unclear and few regulators have been identified. In this study, we developed a reporter that is size excluded from re-compartmentalization following nuclear rupture events. This allows for robust detection of factors influencing nuclear integrity in fixed cells. We combined this with an automated image analysis pipeline in a high-content siRNA screen to identify new proteins that both increase and decrease nuclear rupture frequency in cancer cells. Pathway analysis identified an enrichment of nuclear membrane and ER factors in our hits and we demonstrate that one of these, the protein phosphatase CTDNEP1, is required for nuclear stability. Analysis of known rupture determinants, including an automated quantitative analysis of nuclear lamina gaps, are consistent with CTDNEP1 acting independently of actin and nuclear lamina organization. Our findings provide new insights into the molecular mechanism of nuclear rupture and define a highly adaptable program for rupture analysis that removes a substantial barrier to new discoveries in the field.

Abstract TMP120: Contralesional Corticospinal Tract Reinnervates Cervical Spinal Anatomical Regions With Highest Pre-Lesion Synapse Density After Unilateral M1 Stroke in Mice

Introduction: Stroke leaves many patients with significant motor impairment. Following ischemic lesions of the rodent primary motor cortex, sprouting of the contralesional corticospinal tract (cCST) into the injured hemicord is associated with improved motor recovery. In adult mice, most corticospinal neurons project to spinal interneurons, and few synapse directly on alpha motor neurons. We examined the distribution of newly formed corticospinal synapses in the mouse spinal cord, using genetically encoded tracers, 3D microscopy, and a spinal atlas with machine learning-based automated classification and registration. Hypothesis: We tested the hypothesis that cCST neurons innervate homotopic regions in the injured hemicord after unilateral motor cortex stroke in adult mice. Method: We induced a photothrombotic motor cortex stroke or performed sham surgery (n=4 per group) in 8-11 week old male C57/B6 mice and administered a contralesional motor cortex injection of an anterograde adeno-associated virus expressing both membrane-targeted tdTomato and synaptically-targeted eGFP. Cervical spinal cords were subjected to volumetric imaging via serial two-photon tomography (TissueCyte 1000) to characterize synaptic and axonal density of cCST collaterals in whole cervical spinal cords in both stroke and sham mice. Unbiased, global quantification of axonal and synaptic density across the entire cervical cord was accomplished by the development of a custom automated image analysis pipeline, incorporating a novel 3D spinal cord reference volume, published spinal cord annotations comprising 47 distinct anatomical regions (SpinalJ), and machine learning based pixel classification. Results: We observed cCST synapses in the 6-week post-stroke injured hemicord innervating lamina IV & IX most densely at C7 & C8 (See Figure). Conclusions: Contralesional CST reinnervation of the denervated hemicord is targeted to both lamina and cervical level specific.

High-content screening identifies a small molecule that restores AP-4-dependent protein trafficking in neuronal models of AP-4-associated hereditary spastic paraplegia

Unbiased phenotypic screens in patient-relevant disease models offer the potential to detect therapeutic targets for rare diseases. In this study, we developed a high-throughput screening assay to identify molecules that correct aberrant protein trafficking in adapter protein complex 4 (AP-4) deficiency, a rare but prototypical form of childhood-onset hereditary spastic paraplegia characterized by mislocalization of the autophagy protein ATG9A. Using high-content microscopy and an automated image analysis pipeline, we screened a diversity library of 28,864 small molecules and identified a lead compound, BCH-HSP-C01, that restored ATG9A pathology in multiple disease models, including patient-derived fibroblasts and induced pluripotent stem cell-derived neurons. We used multiparametric orthogonal strategies and integrated transcriptomic and proteomic approaches to delineate potential mechanisms of action of BCH-HSP-C01. Our results define molecular regulators of intracellular ATG9A trafficking and characterize a lead compound for the treatment of AP-4 deficiency, providing important proof-of-concept data for future studies.

Robotics-Driven Manufacturing of Cartilaginous Microtissues for Skeletal Tissue Engineering Applications.

Automated technologies are attractive for enhancing the robust manufacturing of tissue-engineered products for clinical translation. In this work, we present an automation strategy using a robotics platform for media changes, and imaging of cartilaginous microtissues cultured in static microwell platforms. We use an automated image analysis pipeline to extract microtissue displacements and morphological features as noninvasive quality attributes. As a result, empty microwells were identified with a 96% accuracy, and dice coefficient of 0.84 for segmentation. Design of experiment are used for the optimization of liquid handling parameters to minimize empty microwells during long-term differentiation protocols. We found no significant effect of aspiration or dispension speeds at and beyond manual speed. Instead, repeated media changes and time in culture were the driving force or microtissue displacements. As the ovine model is the preclinical model of choice for large skeletal defects, we used ovine periosteum-derived cells to form cartilage-intermediate microtissues. Increased expression of COL2A1 confirms chondrogenic differentiation and RUNX2 shows no osteogenic specification. Histological analysis shows an increased secretion of cartilaginous extracellular matrix and glycosaminoglycans in larger microtissues. Furthermore, microtissue-based implants are capable of forming mineralized tissues and bone after 4 weeks of ectopic implantation in nude mice. We demonstrate the development of an integrated bioprocess for culturing and manipulation of cartilaginous microtissues and anticipate the progressive substitution of manual operations with automated solutions for the manufacturing of microtissue-based living implants.

An automated image analysis pipeline to quantify the coordination and overlap of transcription and replication activity in mammalian genomes.

Transcription-replication conflicts (TRCs) represent a potent endogenous source of replication stress. Besides the spatial and temporal coordination of replication and transcription programs, cells employ many additional mechanisms to resolve TRCs in a timely manner, thereby avoiding replication fork stalling and genomic instability. Proximity ligation assays (PLA) using antibodies against actively elongating RNA Polymerase II (RNAPIIpS2) and PCNA to detect proximity (<40nm) between transcribing RNA polymerases and replication forks can be used to assess and quantify TRC levels in cells. A complementary fluorescence microscopy approach to assess the spatial coordination of transcription and replication activities in the nucleus is to quantify the colocalization (200-400nm) between active transcription and ongoing replication using immunofluorescence staining with an antibody against elongating RNA Polymerase II (RNAPIIpS2) and EdU-Click-it pulse-labelling, respectively. Despite significant efforts to automate image analysis, the need for manual verification, correction, and complementation of automated processes creates a bottleneck for efficient, high-throughput and large-scale imaging. Here, we describe an automated Fiji image analysis macro that allows the user to automate the measurement of RNAPIIpS2 and EdU levels and extract the key parameters such as transcription-replication (TR) colocalization and TRC-PLA foci count from single cells in a high throughput manner. While we showcase the usability of this analysis pipeline for quantifying TR colocalization and TRC-PLA in mouse embryonic stem cells (mESCs), the analysis pipeline is designed as a generally applicable tool allowing the quantification of nuclear signals, colocalization and foci count in various model systems and cell types.

Quantitative photography for rapid, reliable measurement of marine macro‐plastic pollution

Abstract Plastics are now ubiquitous in the environment and have been studied in wildlife and in ecosystems for more than 50 years. Measurement of size, shape and colour data for individual fragments of plastic is labour‐intensive, unreliable and prone to observer bias, particularly when it comes to assessment of colour, which relies on arbitrary and inconsistently defined colour categorisations. There is a clear need for a standard method for data collection on plastic pollution, particularly one that can be readily automated given the number of samples involved. This study describes a new method for standardised photography of marine plastics in the 1–100 mm size range (meso‐ and macro‐plastics), including colour correction to account for any image‐to‐image variation in lighting that may impact colour reproduction or apparent brightness. Automated image analysis is then applied to detect individual fragments of plastic for quantitative measurement of size, shape, and colour. The method was tested on 3793 fragments of debris ingested by Flesh‐footed Shearwaters (Ardenna carneipes) on Lord Howe Island, Australia, and compare results from photos taken in two separate locations using different equipment. Photos were acquired of up to 250 fragments at a time with a spatial resolution of 70 μm/pixel and were colour‐corrected using a reference chart to ensure accurate reproduction of colour. The automated image analysis pipeline was found to have a 98% success rate at detecting fragments, and the different size and shape parameters that can be outputted by the pipeline were compared in terms of usefulness. The evidence shown in this study should strongly encourage the uptake of this method for cataloguing macro‐scale plastic pollution, as it provides substantially higher quality data with accurate, reliable measurements of size, shape and colour for individual plastics that can be readily compared between disparate datasets.

Invasion-Block and S-MARVEL: A high-content screening and image analysis platform identifies ATM kinase as a modulator of melanoma invasion and metastasis.

Robust high-throughput assays are crucial for the effective functioning of a drug discovery pipeline. Herein, we report the development of Invasion-Block, an automated high-content screening platform for measuring invadopodia-mediated matrix degradation as a readout for the invasive capacity of cancer cells. Combined with Smoothen-Mask and Reveal, a custom-designed, automated image analysis pipeline, this platform allowed us to evaluate melanoma cell invasion capacity posttreatment with two libraries of compounds comprising 3840 U.S. Food and Drug Administration (FDA)-approved drugs with well-characterized safety and bioavailability profiles in humans as well as a kinase inhibitor library comprising 210 biologically active compounds. We found that Abl/Src, PKC, PI3K, and Ataxia-telangiectasia mutated (ATM) kinase inhibitors significantly reduced melanoma cell invadopodia formation and cell invasion. Abrogation of ATM expression in melanoma cells via CRISPR-mediated gene knockout reduced 3D invasion in vitro as well as spontaneous lymph node metastasis in vivo. Together, this study established a rapid screening assay coupled with a customized image-analysis pipeline for the identification of antimetastatic drugs. Our study implicates that ATM may serve as a potent therapeutic target for the treatment of melanoma cell spread in patients.

RGB image-based method for phenotyping rust disease progress in pea leaves using R

BackgroundRust is a damaging disease affecting vital crops, including pea, and identifying highly resistant genotypes remains a challenge. Accurate measurement of infection levels in large germplasm collections is crucial for finding new resistance sources. Current evaluation methods rely on visual estimation of disease severity and infection type under field or controlled conditions. While they identify some resistance sources, they are error-prone and time-consuming. An image analysis system proves useful, providing an easy-to-use and affordable way to quickly count and measure rust-induced pustules on pea samples. This study aimed to develop an automated image analysis pipeline for accurately calculating rust disease progression parameters under controlled conditions, ensuring reliable data collection.ResultsA highly efficient and automatic image-based method for assessing rust disease in pea leaves was developed using R. The method’s optimization and validation involved testing different segmentation indices and image resolutions on 600 pea leaflets with rust symptoms. The approach allows automatic estimation of parameters like pustule number, pustule size, leaf area, and percentage of pustule coverage. It reconstructs time series data for each leaf and integrates daily estimates into disease progression parameters, including latency period and area under the disease progression curve. Significant variation in disease responses was observed between genotypes using both visual ratings and image-based analysis. Among assessed segmentation indices, the Normalized Green Red Difference Index (NGRDI) proved fastest, analysing 600 leaflets at 60% resolution in 62 s with parallel processing. Lin’s concordance correlation coefficient between image-based and visual pustule counting showed over 0.98 accuracy at full resolution. While lower resolution slightly reduced accuracy, differences were statistically insignificant for most disease progression parameters, significantly reducing processing time and storage space. NGRDI was optimal at all time points, providing highly accurate estimations with minimal accumulated error.ConclusionsA new image-based method for monitoring pea rust disease in detached leaves, using RGB spectral indices segmentation and pixel value thresholding, improves resolution and precision. It rapidly analyses hundreds of images with accuracy comparable to visual methods and higher than other image-based approaches. This method evaluates rust progression in pea, eliminating rater-induced errors from traditional methods. Implementing this approach to evaluate large germplasm collections will improve our understanding of plant-pathogen interactions and aid future breeding for novel pea cultivars with increased rust resistance.

Emergent coexistence in multispecies microbial communities.

Understanding the mechanisms that maintain microbial biodiversity is a critical aspiration in ecology. Past work on microbial coexistence has largely focused on species pairs, but it is unclear whether pairwise coexistence in isolation is required for coexistence in a multispecies community. To address this question, we conducted hundreds of pairwise competition experiments among the stably coexisting members of 12 different enrichment communities in vitro. To determine the outcomes of these experiments, we developed an automated image analysis pipeline to quantify species abundances. We found that competitive exclusion was the most common outcome, and it was strongly hierarchical and transitive. Because many species that coexist within a stable multispecies community fail to coexist in pairwise co-culture under identical conditions, we concluded that multispecies coexistence is an emergent phenomenon. This work highlights the importance of community context for understanding the origins of coexistence in complex ecosystems.

Single-spore germination analyses reveal that calcium released during Clostridioides difficile germination functions in a feedforward loop.

Clostridioides difficile infections begin when its metabolically dormant spores germinate in response to sensing bile acid germinants alongside amino acid and divalent cation co-germinants in the small intestine. While bile acid germinants are essential for C. difficile spore germination, it is currently unclear whether both co-germinant signals are required. One model proposes that divalent cations, particularly Ca2+, are essential for inducing germination, while another proposes that either co-germinant class can induce germination. The former model is based on the finding that spores defective in releasing large stores of internal Ca2+ in the form of calcium dipicolinic acid (CaDPA) cannot germinate when germination is induced with bile acid germinant and amino acid co-germinant alone. However, since the reduced optical density of CaDPA-less spores makes it difficult to accurately measure their germination, we developed a novel automated, time-lapse microscopy-based germination assay to analyze CaDPA mutant germination at the single-spore level. Using this assay, we found that CaDPA mutant spores germinate in the presence of amino acid co-germinant and bile acid germinant. Higher levels of amino acid co-germinants are nevertheless required to induce CaDPA mutant spores to germinate relative to WT spores because CaDPA released by WT spores during germination can function in a feedforward loop to potentiate the germination of other spores within the population. Collectively, these data indicate that Ca2+ is not essential for inducing C. difficile spore germination because amino acid and Ca2+ co-germinant signals are sensed by parallel signaling pathways. IMPORTANCE Clostridioides difficile spore germination is essential for this major nosocomial pathogen to initiate infection. C. difficile spores germinate in response to sensing bile acid germinant signals alongside co-germinant signals. There are two classes of co-germinant signals: Ca2+ and amino acids. Prior work suggested that Ca2+ is essential for C. difficile spore germination based on bulk population analyses of germinating CaDPA mutant spores. Since these assays rely on optical density to measure spore germination and the optical density of CaDPA mutant spores is reduced relative to WT spores, this bulk assay is limited in its capacity to analyze germination. To overcome this limitation, we developed an automated image analysis pipeline to monitor C. difficile spore germination using time-lapse microscopy. With this analysis pipeline, we demonstrate that, although Ca2+ is dispensable for inducing C. difficile spore germination, CaDPA can function in a feedforward loop to potentiate the germination of neighboring spores.

Fully Automatic initialization and segmentation of left and right ventricles for large-scale cardiac MRI using a deeply supervised network and 3D-ASM

Background and objectiveThe sheer volume of data generated by population imaging studies is unparalleled by current capabilities to extract objective and quantitative cardiac phenotypes; subjective and time-consuming manual image analysis remains the gold standard. Automated image analytics to compute quantitative imaging biomarkers of cardiac function are desperately needed. Data volumes and their variability pose a challenge to most state-of-the-art methods for endo- and epicardial contours, which lack robustness when applied to very large datasets. Our aim is to develop an analysis pipeline for the automatic quantification of cardiac function from cine magnetic resonance imaging data. MethodThis work adopt 4,638 cardiac MRI cases coming from UK Biobank with ground truth available for left and RV contours. A hybrid and robust algorithm is proposed to improve the accuracy of automatic left and right ventricle segmentation by harnessing the localization accuracy of deep learning and the morphological accuracy of 3D-ASM (three-dimensional active shape models). The contributions of this paper are three-fold. First, a fully automatic method is proposed for left and right ventricle initialization and cardiac MRI segmentation by taking full advantage of spatiotemporal constraint. Second, a deeply supervised network is introduced to train and segment the heart. Third, the 3D-ASM image search procedure is improved by combining image intensity models with convolutional neural network (CNN) derived distance maps improving endo- and epicardial edge localization. ResultsThe proposed architecture outperformed the state of the art for cardiac MRI segmentation from UK Biobank. The statistics of RV landmarks detection errors for Triscuspid valve and RV apex are 4.17 mm and 5.58 mm separately. The overlap metric, mean contour distance, Hausdorff distance and cardiac functional parameters are calculated for the LV (Left Ventricle) and RV (Right Ventricle) contour segmentation. Bland–Altman analysis for clinical parameters shows that the results from our automated image analysis pipelines are in good agreement with results from expert manual analysis. ConclusionsOur hybrid scheme combines deep learning and statistical shape modeling for automatic segmentation of the LV/RV from cardiac MRI datasets is effective and robust and can compute cardiac functional indexes from population imaging.

FiNuTyper: Design and validation of an automated deep learning-based platform for simultaneous fiber and nucleus type analysis in human skeletal muscle.

While manual quantification is still considered the gold standard for skeletal muscle histological analysis, it is time-consuming and prone to investigator bias. To address this challenge, we assembled an automated image analysis pipeline, FiNuTyper (Fiber and Nucleus Typer). We integrated recently developed deep learning-based image segmentation methods, optimized for unbiased evaluation of fresh and postmortem human skeletal muscle, and utilized SERCA1 and SERCA2 as type-specific myonucleus and myofiber markers after validating them against the traditional use of MyHC isoforms. Parameters including cross-sectional area, myonuclei per fiber, myonuclear domain, central myonuclei per fiber, and grouped myofiber ratio were determined in a fiber-type-specific manner, revealing that a large degree of sex- and muscle-related heterogeneity could be detected using the pipeline. Our platform was also tested on pathological muscle tissue (ALS and IBM) and adapted for the detection of other resident cell types (leucocytes, satellite cells, capillary endothelium). In summary, we present an automated image analysis tool for the simultaneous quantification of myofiber and myonuclear types, to characterize the composition and structure of healthy and diseased human skeletal muscle.

Abstract P2-09-03: Quantification of HER2 expression and spatial biology using computational pathology: A cross-assay validation study in breast cancer

Abstract Background Conventional pathologic scoring of HER2 by IHC is proven to distinguish potential responders to trastuzumab but has not been effective for next generation antibody drug conjugates (ADCs) such as trastuzumab deruxtecan (T-DXd), which is capable of bystander killing. Several alternative approaches have been deployed to measure HER2, including immunofluorescence and mRNA sequencing. We have developed a novel and fully automated computational pathology technique, Quantitative Continuous Scoring (QCS), to quantify the level and distribution of HER2 from digitized HER2 IHC slides in an objective, quantifiable, and reproducible manner on a per-cell basis [Gustavson et al., SABCS 2020]. To further validate this approach, we performed a systematic multi-omic comparison of QCS to orthogonal methods of HER2 quantitation on a cohort of primary and metastatic breast cancer cases (N=30). Methods HER2 was evaluated using three independent methods on serial tissue sections obtained from 30 archival FFPE breast cancer samples distributed over the full range of HER2 expression, from 0 to 3+. HER2-IHC staining (clone 4B5, Roche Tissue Diagnostics) was performed using standard methods and cases were scored by two pathologists using CAP/ASCO guidelines and H-scores were assigned. We performed FISH (HER2 IQFISH pharmDx [Dako]; PathVysion HER-2 DNA Probe Kit [Vysis]), mRNA quantification of ERBB2 transcript levels (Nano String), and immunofluorescence (IF; HER2 clone 29D8, CST). Imaged with Vectra (Akoya) and analyzed with Halo (Indica). QCS readouts were generated from the above-mentioned digital images of IHC slides by using a fully automated image analysis pipeline; readouts included per-cell staining intensity measurements of membranes and cytoplasmic sub-compartments in terms of optical density (OD) [Van der Laak, JQCS 2000], which were aggregated to a single slide-level score. Additionally, using the OD measurements and the cell locations, a Spatial Proximity Score (SPS) was computed, summing the percentage of cells with OD≥10 (corresponding to the limit of visual detection of IHC staining) as well as the percentage of cells with OD&amp;lt; 10 within a prespecified radius (25µm) of a neighboring cell with OD≥10. Results Our analysis demonstrated that QCS-based scoring correlates with orthogonal measurements used in this study. Comparing protein-based assays, the observed Pearson correlation was R=0.88 between QCS median membrane OD and IHC H-scores, R=0.86 with IF-based HER2 mean cell expression intensity, and R=0.85 with IF-based H-scores. Correlation with transcriptomic profiling was R=0.81 for OD vs. mRNA, however ERBB2 transcript levels did not distinguish between HER2 0, 1+, and 2+ FISH negative cases, while QCS was able to do so. Correlation between protein-based and nucleic-acid based assays were numerically worse, with R=0.64 for OD vs. FISH. All samples (including those with HER2 IHC scores of 0 and H-Scores &amp;lt; 10) had at least ~20% of cells with quantifiable HER2 expression by OD, the presence of which was confirmed using IF. For cases in the lowest quartile of HER2 expression by OD, SPS identified 20-50% additional HER2-null cells that were in close proximity to HER2-expressing cells that may be vulnerable to bystander killing. Conclusion QCS-based scoring is consistent with orthogonal protein-based measurements across the range of HER2 expression. Most importantly, QCS derived-spatial analysis features identify additional patients in the lower end of HER2 expression that might be highly relevant for ADC response prediction, particularly if a drug exerts bystander activity. Further clinical verification and validation on large cohorts is needed. Footnote: This study was approved by the IRB at MSKCC. Citation Format: Joshua Drago, Zonera Hassan, Jan Zaucha, Ansh Kapil, Fatemeh Derakhshan, Fresia Pareja, Shimulov Anatoliy, Fanni Ratzon, Travis J Hollman, Claire Myers, Jessica Chan, Andrea Spitzmuller, Mark Gustavson, Danielle Carroll, Dara Ross, Jorge Reis-Filho, Carl barrett, Sihem Khalifa, Schmidt Guenter, Hadassah Sade, Sarat Chandarlapaty. Quantification of HER2 expression and spatial biology using computational pathology: A cross-assay validation study in breast cancer [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P2-09-03.

Automated Image Analysis Pipeline Development to Monitor Disease Progression in Muscular Dystrophy Using Cell Profiler.

Muscular dystrophies are inherited disorders that are characterized by progressive muscle degeneration. These disorders are caused by mutations in the genes encoding structural elements within the muscle, which leads to increased vulnerability to mechanical stress and sarcolemma damage. Although myofibers have the capacity to regenerate, the newly formed myofibers still harbor genetic mutation, which induces continuous cycles of muscle fiber death and regeneration. This repeated cycling is accompanied by an inflammatory response which eventually provokes excessive fibrotic deposition. The histopathological changes in skeletal muscle tissue are central to the disease pathogenesis. Analysis of muscle histopathology is the gold standard for monitoring muscle health and disease progression. However, manual, or semi-manual quantification methods, are not only immensely tedious but can be subjective. Here, we present four image analysis pipelines built in CellProfiler which enable users without a background in computer vision or programming to quantitatively analyze biological images. These image analysis pipelines are designed to quantify skeletal muscle histopathological staining for membrane damage, the abundance and size distribution of regenerating muscle fibers, inflammation via quantification of CD68+ M1 macrophages, and collagen deposition. Additionally, we discuss methods to address common errors associated with the quantification of microscopy images. These automated tools can not only improve workflow efficiency but can provide a better understanding of the histopathological progression of muscular dystrophy.

N-terminal mutant huntingtin deposition correlates with CAG repeat length and symptom onset, but not neuronal loss in Huntington's disease

Huntington's disease (HD) is caused by a CAG repeat expansion mutation in the gene encoding the huntingtin (Htt) protein, with mutant Htt protein subsequently forming aggregates within the brain. Mutant Htt is a current target for novel therapeutic strategies for HD, however, the lack of translation from preclinical research to disease-modifying treatments highlights the need to improve our understanding of the role of Htt protein in the human brain. This study aims to undertake an immunohistochemical screen of 12 candidate antibodies against various sequences along the Htt protein to characterize Htt distribution and expression in post-mortem human brain tissue microarrays (TMAs).Immunohistochemistry was performed on middle temporal gyrus TMAs comprising of up to 28 HD and 27 age-matched control cases, using 12 antibodies specific to various sequences along the Htt protein. From this study, six antibodies directed to the Htt N-terminus successfully immunolabeled human brain tissue. Htt aggregates and Htt protein expression levels for the six successful antibodies were subsequently quantified with a customized automated image analysis pipeline on the TMAs. A 2.5–12 fold increase in the number of Htt aggregates were detected in HD cases using antibodies MAB5374, MW1, and EPR5526, despite no change in overall Htt protein expression compared to control cases, suggesting a redistribution of Htt into aggregates in HD. MAB5374, MW1, and EPR5526 Htt aggregate numbers were positively correlated with CAG repeat length, and negatively correlated with the age of symptom onset in HD. However, the number of Htt aggregates did not correlate with the degree of striatal degeneration or the degree of cortical neuron loss. Together, these results suggest that longer CAG repeat lengths correlate with Htt aggregation in the HD human brain, and greater Htt cortical aggregate deposition is associated with an earlier age of symptom onset in HD. This study also reinforces that antibodies MAB5492, MW8, and 2B7 which have been utilized to characterize Htt in animal models of HD do not specifically immunolabel Htt aggregates in HD human brain tissue exclusively, thereby highlighting the need for validated means of Htt detection to support drug development for HD.

Abstract 3379: Multi-cohort study shows HCC is associated with a relative regional reduction in TGFBR2 in liver biopsies from patients with cirrhosis

Abstract Background: Early detection of hepatocellular carcinoma (HCC) is challenging partly because most patients also have cirrhosis. Furthermore, patients with cirrhosis are at high risk of developing HCC. Loss of function of the transforming growth factor β (TGF-β) pathway is associated with HCC. The goal of this study was to determine if a reduction in TGF-β receptors (TGFBR1, TGFBR2) was able to reliably and reproducibly predict HCC in cirrhotic tissue. Methods: We performed quantitative immunohistochemical (IHC) analysis of biopsy tissue from patients with cirrhosis with or without HCC using 3 separate sets of samples. We developed an automated image analysis pipeline to quantify IHC of TGFBR1 and TGFRB2 in liver biopsies. Using logistic regression modeling, we evaluated the predictive power of staining intensity of TGFBR1, TGFBR2, or the combination in differentiating HCC from cirrhosis. Results: We found a consistently significant reduction in TGFBR2 in HCC (p &amp;lt; 0.02) among independent patient cohorts from 3 separate institutions. With the data, we developed an artificial intelligence-based image analysis pipeline that confirmed the significant reduction in TGFBR2 staining intensity in HCC compared to cirrhotic tissue (p &amp;lt; 0.002). Unfortunately, logistic regression modeling of the quantitative data showed that there is not a simple threshold of TGFBR2 staining intensity that is diagnostic of HCC. Instead, in 84% of samples with HCC, HCC was associated with a relative regional reduction in TGFBR2 compared to tumor-adjacent cirrhotic tissue. Conclusions: The qualitative and quantitative analyses indicate that evaluating biopsy samples for a reduction in TGFBR2 abundance relative to the surrounding cirrhotic tissue could be an early indicator of the development of HCC in high-risk groups undergoing surveillance for HCC. Furthermore, an automated image analysis pipeline for quantitative IHC of TGFBR staining could improve the reliability of image analysis across institutions participating in clinical trials. Citation Format: Krishanu Bhowmick, Richard Amdur, Xiyan Xiang, Herbert Yu, Linda L. Wong, Shuyun Rao, Aiwu R. He, Karan Amin, Daewa Zaheer, Raj K. Narayan, Sanjaya K. Satapathy, Chandan Guha, Patricia S. Latham, Kirti Shetty, Nancy R. Gough, Lopa Mishra. Multi-cohort study shows HCC is associated with a relative regional reduction in TGFBR2 in liver biopsies from patients with cirrhosis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3379.