Searcharxiv⌕ Search

arXiv subjects

Anirban Mukhopadhyay

Publications and source records attributed to Anirban Mukhopadhyay.

At least 55 records · Page 3Linked to original sources

FrOoDo: Framework for Out-of-Distribution Detection

FrOoDo is an easy-to-use and flexible framework for Out-of-Distribution detection tasks in digital pathology. It can be used with PyTorch classification and segmentation models, and its modular design allows for easy extension. The goal is to automate the task of OoD Evaluation such that research can focus on the main goal of either designing new models, new methods or evaluating a new dataset. The code can be found at https://github.com/MECLabTUDA/FrOoDo.

cs.CV↗

OSINT Research Studios: A Flexible Crowdsourcing Framework to Scale Up Open Source Intelligence Investigations

Open Source Intelligence (OSINT) investigations, which rely entirely on publicly available data such as social media, play an increasingly important role in solving crimes and holding governments accountable. The growing volume of data and complex nature of tasks, however, means there is a pressing need to scale and speed up OSINT investigations. Expert-led crowdsourcing approaches show promise but tend to either focus on narrow tasks or domains or require resource-intense, long-term relationships between expert investigators and crowds. We address this gap by providing a flexible framework that enables investigators across domains to enlist crowdsourced support for the discovery and verification of OSINT. We use a design-based research (DBR) approach to develop OSINT Research Studios (ORS), a sociotechnical system in which novice crowds are trained to support professional investigators with complex OSINT investigations. Through our qualitative evaluation, we found that ORS facilitates ethical and effective OSINT investigations across multiple domains. We also discuss broader implications of expert-crowd collaboration and opportunities for future work.

cs.HC↗

Continual atlas-based segmentation of prostate MRI

Continual learning (CL) methods designed for natural image classification often fail to reach basic quality standards for medical image segmentation. Atlas-based segmentation, a well-established approach in medical imaging, incorporates domain knowledge on the region of interest, leading to semantically coherent predictions. This is especially promising for CL, as it allows us to leverage structural information and strike an optimal balance between model rigidity and plasticity over time. When combined with privacy-preserving prototypes, this process offers the advantages of rehearsal-based CL without compromising patient privacy. We propose Atlas Replay, an atlas-based segmentation approach that uses prototypes to generate high-quality segmentation masks through image registration that maintain consistency even as the training distribution changes. We explore how our proposed method performs compared to state-of-the-art CL methods in terms of knowledge transferability across seven publicly available prostate segmentation datasets. Prostate segmentation plays a vital role in diagnosing prostate cancer, however, it poses challenges due to substantial anatomical variations, benign structural differences in older age groups, and fluctuating acquisition parameters. Our results show that Atlas Replay is both robust and generalizes well to yet-unseen domains while being able to maintain knowledge, unlike end-to-end segmentation methods. Our code base is available under https://github.com/MECLabTUDA/Atlas-Replay.

cs.CV↗

From Pointwise to Powerhouse: Initialising Neural Networks with Generative Models

Traditional initialisation methods, e.g. He and Xavier, have been effective in avoiding the problem of vanishing or exploding gradients in neural networks. However, they only use simple pointwise distributions, which model one-dimensional variables. Moreover, they ignore most information about the architecture and disregard past training experiences. These limitations can be overcome by employing generative models for initialisation. In this paper, we introduce two groups of new initialisation methods. First, we locally initialise weight groups by employing variational autoencoders. Secondly, we globally initialise full weight sets by employing graph hypernetworks. We thoroughly evaluate the impact of the employed generative models on state-of-the-art neural networks in terms of accuracy, convergence speed and ensembling. Our results show that global initialisations result in higher accuracy and faster initial convergence speed. However, the implementation through graph hypernetworks leads to diminished ensemble performance on out of distribution data. To counteract, we propose a modification called noise graph hypernetwork, which encourages diversity in the produced ensemble members. Furthermore, our approach might be able to transfer learned knowledge to different image distributions. Our work provides insights into the potential, the trade-offs and possible modifications of these new initialisation methods.

cs.CV↗

Exploring SAM Ablations for Enhancing Medical Segmentation in Radiology and Pathology

Medical imaging plays a critical role in the diagnosis and treatment planning of various medical conditions, with radiology and pathology heavily reliant on precise image segmentation. The Segment Anything Model (SAM) has emerged as a promising framework for addressing segmentation challenges across different domains. In this white paper, we delve into SAM, breaking down its fundamental components and uncovering the intricate interactions between them. We also explore the fine-tuning of SAM and assess its profound impact on the accuracy and reliability of segmentation results, focusing on applications in radiology (specifically, brain tumor segmentation) and pathology (specifically, breast cancer segmentation). Through a series of carefully designed experiments, we analyze SAM's potential application in the field of medical imaging. We aim to bridge the gap between advanced segmentation techniques and the demanding requirements of healthcare, shedding light on SAM's transformative capabilities.

cs.CV↗

Statistics of Moduli Space of vector bundles II

Let $X$ be a smooth irreducible projective curve of genus $g \geq 2$ over a finite field $\F_{q}$ of characteristic $p$ with $q$ elements such that the function field $\F_{q}(X)$ is a geometric Galois extension of the rational function field of degree $N.$ Consider $gcd(n,d)=1$, let $M_{L}(n,d)$ be the moduli space of rank $n$ stable vector bundles over $X$ with fixed determinant isomorphic to a $\mathbb F_q$-rational line bundle $L$. Suppose $N_q (M_L(n,d))$ denotes the cardinality of the set of $\F_{q}$-rational points of $M_{L}(n,d)$. We give an asymptotic bound of $\log(N_{q}(M_{L}(n,d)) - (n^2-1)(g-1)\log{q})$ for large genus $g,$ depending on $N$. Further, considering this logarithmic difference as a random variable, we prove a central limit theorem over a large family of hyperelliptic curves with uniform probability measure. Further, over the same family of hyperelliptic curves, we study the distribution of $\F_{q}$-rational points over the moduli space of rank $2$ stable vector bundles with trivial determinant $M^{s}_{\mathcal{O}_{H}}(2,0)$ and it's Seshadri desingularisation ${\widetilde{N}}$ by choosing an appropriate random variable in each case. We also see that the corresponding random variables having standard Gaussian distribution as $g$ and $q$ tends to infinity.

math.AG↗

M3D-NCA: Robust 3D Segmentation with Built-in Quality Control

Medical image segmentation relies heavily on large-scale deep learning models, such as UNet-based architectures. However, the real-world utility of such models is limited by their high computational requirements, which makes them impractical for resource-constrained environments such as primary care facilities and conflict zones. Furthermore, shifts in the imaging domain can render these models ineffective and even compromise patient safety if such errors go undetected. To address these challenges, we propose M3D-NCA, a novel methodology that leverages Neural Cellular Automata (NCA) segmentation for 3D medical images using n-level patchification. Moreover, we exploit the variance in M3D-NCA to develop a novel quality metric which can automatically detect errors in the segmentation process of NCAs. M3D-NCA outperforms the two magnitudes larger UNet models in hippocampus and prostate segmentation by 2% Dice and can be run on a Raspberry Pi 4 Model B (2GB RAM). This highlights the potential of M3D-NCA as an effective and efficient alternative for medical image segmentation in resource-constrained environments.

cs.CV↗

Jointly Exploring Client Drift and Catastrophic Forgetting in Dynamic Learning

Federated and Continual Learning have emerged as potential paradigms for the robust and privacy-aware use of Deep Learning in dynamic environments. However, Client Drift and Catastrophic Forgetting are fundamental obstacles to guaranteeing consistent performance. Existing work only addresses these problems separately, which neglects the fact that the root cause behind both forms of performance deterioration is connected. We propose a unified analysis framework for building a controlled test environment for Client Drift -- by perturbing a defined ratio of clients -- and Catastrophic Forgetting -- by shifting all clients with a particular strength. Our framework further leverages this new combined analysis by generating a 3D landscape of the combined performance impact from both. We demonstrate that the performance drop through Client Drift, caused by a certain share of shifted clients, is correlated to the drop from Catastrophic Forgetting resulting from a corresponding shift strength. Correlation tests between both problems for Computer Vision (CelebA) and Medical Imaging (PESO) support this new perspective, with an average Pearson rank correlation coefficient of over 0.94. Our framework's novel ability of combined spatio-temporal shift analysis allows us to investigate how both forms of distribution shift behave in mixed scenarios, opening a new pathway for better generalization. We show that a combination of moderate Client Drift and Catastrophic Forgetting can even improve the performance of the resulting model (causing a "Generalization Bump") compared to when only one of the shifts occurs individually. We apply a simple and commonly used method from Continual Learning in the federated setting and observe this phenomenon to be reoccurring, leveraging the ability of our framework to analyze existing and novel methods for Federated and Continual Learning.

cs.LG↗

On higher dimensional Poissonian pair correlation

In this article we study the pair correlation statistic for higher dimensional sequences. We show that for any $d\geq 2$, strictly increasing sequences $(a_n^{(1)}),\ldots, (a_n^{(d)})$ of natural numbers have metric Poissonian pair correlation with respect to sup-norm if their joint additive energy is $O(N^{3-δ})$ for any $δ>0$. Further, in two dimension, we establish an analogous result with respect to $2$-norm. As a consequence, it follows that $(\{nα\}, \{n^2β\})$ and $(\{nα\}, \{[n\log^An]β\})$ ($A \in [1,2]$) have Poissonian pair correlation for almost all $(α,β)\in \mathbb{R}^2$ with respect to sup-norm and $2$-norm. This gives a negative answer to the question raised by Hofer and Kaltenböck [15]. The proof uses estimates for 'Generalized' GCD-sums.

math.NT↗

Synthesising Rare Cataract Surgery Samples with Guided Diffusion Models

Cataract surgery is a frequently performed procedure that demands automation and advanced assistance systems. However, gathering and annotating data for training such systems is resource intensive. The publicly available data also comprises severe imbalances inherent to the surgical process. Motivated by this, we analyse cataract surgery video data for the worst-performing phases of a pre-trained downstream tool classifier. The analysis demonstrates that imbalances deteriorate the classifier's performance on underrepresented cases. To address this challenge, we utilise a conditional generative model based on Denoising Diffusion Implicit Models (DDIM) and Classifier-Free Guidance (CFG). Our model can synthesise diverse, high-quality examples based on complex multi-class multi-label conditions, such as surgical phases and combinations of surgical tools. We affirm that the synthesised samples display tools that the classifier recognises. These samples are hard to differentiate from real images, even for clinical experts with more than five years of experience. Further, our synthetically extended data can improve the data sparsity problem for the downstream task of tool classification. The evaluations demonstrate that the model can generate valuable unseen examples, allowing the tool classifier to improve by up to 10% for rare cases. Overall, our approach can facilitate the development of automated assistance systems for cataract surgery by providing a reliable source of realistic synthetic data, which we make available for everyone.

eess.IV↗

CoSINT: Designing a Collaborative Capture the Flag Competition to Investigate Misinformation

Crowdsourced investigations shore up democratic institutions by debunking misinformation and uncovering human rights abuses. However, current crowdsourcing approaches rely on simplistic collaborative or competitive models and lack technological support, limiting their collective impact. Prior research has shown that blending elements of competition and collaboration can lead to greater performance and creativity, but crowdsourced investigations pose unique analytical and ethical challenges. In this paper, we employed a four-month-long Research through Design process to design and evaluate a novel interaction style called collaborative capture the flag competitions (CoCTFs). We instantiated this interaction style through CoSINT, a platform that enables a trained crowd to work with professional investigators to identify and investigate social media misinformation. Our mixed-methods evaluation showed that CoSINT leverages the complementary strengths of competition and collaboration, allowing a crowd to quickly identify and debunk misinformation. We also highlight tensions between competition versus collaboration and discuss implications for the design of crowdsourced investigations.

cs.HC↗

GaNDLF: A Generally Nuanced Deep Learning Framework for Scalable End-to-End Clinical Workflows in Medical Imaging

Deep Learning (DL) has the potential to optimize machine learning in both the scientific and clinical communities. However, greater expertise is required to develop DL algorithms, and the variability of implementations hinders their reproducibility, translation, and deployment. Here we present the community-driven Generally Nuanced Deep Learning Framework (GaNDLF), with the goal of lowering these barriers. GaNDLF makes the mechanism of DL development, training, and inference more stable, reproducible, interpretable, and scalable, without requiring an extensive technical background. GaNDLF aims to provide an end-to-end solution for all DL-related tasks in computational precision medicine. We demonstrate the ability of GaNDLF to analyze both radiology and histology images, with built-in support for k-fold cross-validation, data augmentation, multiple modalities and output classes. Our quantitative performance evaluation on numerous use cases, anatomies, and computational tasks supports GaNDLF as a robust application framework for deployment in clinical workflows.

cs.LG↗

What is Wrong with Continual Learning in Medical Image Segmentation?

Continual learning protocols are attracting increasing attention from the medical imaging community. In continual environments, datasets acquired under different conditions arrive sequentially; and each is only available for a limited period of time. Given the inherent privacy risks associated with medical data, this setup reflects the reality of deployment for deep learning diagnostic radiology systems. Many techniques exist to learn continuously for image classification, and several have been adapted to semantic segmentation. Yet most struggle to accumulate knowledge in a meaningful manner. Instead, they focus on preventing the problem of catastrophic forgetting, even when this reduces model plasticity and thereon burdens the training process. This puts into question whether the additional overhead of knowledge preservation is worth it - particularly for medical image segmentation, where computation requirements are already high - or if maintaining separate models would be a better solution. We propose UNEG, a simple and widely applicable multi-model benchmark that maintains separate segmentation and autoencoder networks for each training stage. The autoencoder is built from the same architecture as the segmentation network, which in our case is a full-resolution nnU-Net, to bypass any additional design decisions. During inference, the reconstruction error is used to select the most appropriate segmenter for each test image. Open this concept, we develop a fair evaluation scheme for different continual learning settings that moves beyond the prevention of catastrophic forgetting. Our results across three regions of interest (prostate, hippocampus, and right ventricle) show that UNEG outperforms several continual learning methods, reinforcing the need for strong baselines in continual learning research.

cs.CV↗

Med-NCA: Robust and Lightweight Segmentation with Neural Cellular Automata

Access to the proper infrastructure is critical when performing medical image segmentation with Deep Learning. This requirement makes it difficult to run state-of-the-art segmentation models in resource-constrained scenarios like primary care facilities in rural areas and during crises. The recently emerging field of Neural Cellular Automata (NCA) has shown that locally interacting one-cell models can achieve competitive results in tasks such as image generation or segmentations in low-resolution inputs. However, they are constrained by high VRAM requirements and the difficulty of reaching convergence for high-resolution images. To counteract these limitations we propose Med-NCA, an end-to-end NCA training pipeline for high-resolution image segmentation. Our method follows a two-step process. Global knowledge is first communicated between cells across the downscaled image. Following that, patch-based segmentation is performed. Our proposed Med-NCA outperforms the classic UNet by 2% and 3% Dice for hippocampus and prostate segmentation, respectively, while also being 500 times smaller. We also show that Med-NCA is by design invariant with respect to image scale, shape and translation, experiencing only slight performance degradation even with strong shifts; and is robust against MRI acquisition artefacts. Med-NCA enables high-resolution medical image segmentation even on a Raspberry Pi B+, arguably the smallest device able to run PyTorch and that can be powered by a standard power bank.

eess.IV↗

The state-of-the-art 3D anisotropic intracranial hemorrhage segmentation on non-contrast head CT: The INSTANCE challenge

Automatic intracranial hemorrhage segmentation in 3D non-contrast head CT (NCCT) scans is significant in clinical practice. Existing hemorrhage segmentation methods usually ignores the anisotropic nature of the NCCT, and are evaluated on different in-house datasets with distinct metrics, making it highly challenging to improve segmentation performance and perform objective comparisons among different methods. The INSTANCE 2022 was a grand challenge held in conjunction with the 2022 International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI). It is intended to resolve the above-mentioned problems and promote the development of both intracranial hemorrhage segmentation and anisotropic data processing. The INSTANCE released a training set of 100 cases with ground-truth and a validation set with 30 cases without ground-truth labels that were available to the participants. A held-out testing set with 70 cases is utilized for the final evaluation and ranking. The methods from different participants are ranked based on four metrics, including Dice Similarity Coefficient (DSC), Hausdorff Distance (HD), Relative Volume Difference (RVD) and Normalized Surface Dice (NSD). A total of 13 teams submitted distinct solutions to resolve the challenges, making several baseline models, pre-processing strategies and anisotropic data processing techniques available to future researchers. The winner method achieved an average DSC of 0.6925, demonstrating a significant growth over our proposed baseline method. To the best of our knowledge, the proposed INSTANCE challenge releases the first intracranial hemorrhage segmentation benchmark, and is also the first challenge that intended to resolve the anisotropic problem in 3D medical image segmentation, which provides new alternatives in these research fields.

eess.IV↗

Federated Stain Normalization for Computational Pathology

Although deep federated learning has received much attention in recent years, progress has been made mainly in the context of natural images and barely for computational pathology. However, deep federated learning is an opportunity to create datasets that reflect the data diversity of many laboratories. Further, the effort of dataset construction can be divided among many. Unfortunately, existing algorithms cannot be easily applied to computational pathology since previous work presupposes that data distributions of laboratories must be similar. This is an unlikely assumption, mainly since different laboratories have different staining styles. As a solution, we propose BottleGAN, a generative model that can computationally align the staining styles of many laboratories and can be trained in a privacy-preserving manner to foster federated learning in computational pathology. We construct a heterogenic multi-institutional dataset based on the PESO segmentation dataset and improve the IOU by 42\% compared to existing federated learning algorithms. An implementation of BottleGAN is available at https://github.com/MECLabTUDA/BottleGAN

eess.IV↗

Detecting respiratory motion artefacts for cardiovascular MRIs to ensure high-quality segmentation

While machine learning approaches perform well on their training domain, they generally tend to fail in a real-world application. In cardiovascular magnetic resonance imaging (CMR), respiratory motion represents a major challenge in terms of acquisition quality and therefore subsequent analysis and final diagnosis. We present a workflow which predicts a severity score for respiratory motion in CMR for the CMRxMotion challenge 2022. This is an important tool for technicians to immediately provide feedback on the CMR quality during acquisition, as poor-quality images can directly be re-acquired while the patient is still available in the vicinity. Thus, our method ensures that the acquired CMR holds up to a specific quality standard before it is used for further diagnosis. Therefore, it enables an efficient base for proper diagnosis without having time and cost-intensive re-acquisitions in cases of severe motion artefacts. Combined with our segmentation model, this can help cardiologists and technicians in their daily routine by providing a complete pipeline to guarantee proper quality assessment and genuine segmentations for cardiovascular scans. The code base is available at https://github.com/MECLabTUDA/QA_med_data/tree/dev_QA_CMRxMotion.

eess.IV↗

Correlation of multiplicative functions over function fields

In this article we study the asymptotic behaviour of the correlation functions over polynomial ring $\mathbb{F}_q[x]$. Let $\mathcal{M}_{n, q}$ and $\mathcal{P}_{n, q}$ be the set of all monic polynomials and monic irreducible polynomials of degree $n$ over $\mathbb{F}_q$ respectively. For multiplicative functions $ψ_1$ and $ψ_2$ on $\mathbb{F}_q[x]$, we obtain asymptotic formula for the following correlation functions for a fixed $q$ and $n\to \infty$ \begin{align*} &S_{2}(n, q):=\displaystyle\sum_{f\in \mathcal{M}_{n, q}}ψ_1(f+h_1) ψ_2(f+h_2), \\ &R_2(n, q):=\displaystyle\sum_{P\in \mathcal{P}_{n, q}}ψ_1(P+h_1)ψ_2(P+h_2), \end{align*} where $h_1, h_2$ are fixed polynomials of degree $<n$ over $\mathbb{F}_q$. As a consequence, for real valued additive functions $\tilde{ψ_1}$ and $\tilde{ψ_2}$ on $\mathbb{F}_q[x]$ we show that for a fixed $q$ and $n\to \infty$, the following distribution functions \begin{align*} &\frac{1}{|\mathcal{M}_{n, q}|}\Big|\{f\in \mathcal{M}_{n, q} : \tilde{ψ_1}(f+h_1)+\tilde{ψ_2}(f+h_2)\leq x\}\Big|,\\ & \frac{1}{|\mathcal{P}_{n, q}|}\Big|\{P\in \mathcal{P}_{n, q} : \tilde{ψ_1}(P+h_1)+\tilde{ψ_2}(P+h_2)\leq x\}\Big| \end{align*} converges weakly towards a limit distribution.

math.NT↗