arXiv subjects

Sourav Bhattacharjee

Publications and source records attributed to Sourav Bhattacharjee.

At least 19 recordsLinked to original sources

arXiv2026

Beyond Discrimination: Evaluating Probability Calibration in Radiomics-Image Fusion for Renal Cell Carcinoma Classification

Reliable probability estimates are important for the clinical interpretation of machine-learning classifiers, yet radiomics--image fusion models for renal cell carcinoma (RCC) are commonly evaluated primarily by discrimination. Using 396 patients from the KiTS23 dataset, we evaluated whether adding radiomics to four image-model families improved probabilistic performance in addition to discrimination. Seventeen individual models and four probability-average ensembles were assessed using discrimination, probability calibration, and proper scoring measures. Across 12 matched image-only versus fusion comparisons, AUC and AP improved in all 12, while Brier score and negative log likelihood (NLL) improved in 11 of 12 and expected calibration error (ECE) in eight of 12. The 3D MedVAE gated-fusion model achieved the strongest overall individual-model profile. Probability averaging across concatenation, cross-attention, and gated fusion produced favourable point estimates across all five primary metrics relative to image-only models in all four families, although paired-bootstrap support was stronger for Brier score and NLL than for ECE. These findings show that radiomics--image fusion can improve probabilistic prediction quality beyond discrimination, while calibration behaviour varies across image architectures and fusion strategies. We show that our proposed probability-average ensemble models improve discrimination and calibration simultaneously across all architectures in the study.

eess.IV↗

arXiv2026

Beyond Gate Weights: Decision-Level Interpretation of Radiomics-Image Feature Fusion for RCC Classification

Gate coefficients describe how a fusion layer weights intermediate features, but they do not necessarily reflect the magnitude of branch contributions at the classifier output. We investigate this distinction in a fixed 3D MedVAE--radiomics model for clear cell versus non-clear cell renal cell carcinoma classification from contrast-enhanced CT. Because the model uses a linear classification head, the binary decision logit can be decomposed exactly into a gated image term, a direct radiomics term, and a bias. We define effective radiomics dominance as the direct radiomics term's share of the summed absolute non-bias branch contributions and compare it with structural radiomics weighting. Internally, mean structural radiomics weighting was 0.823, whereas mean effective radiomics dominance was 0.148, with a modest case-wise association (\r{ho} = 0.290). The gated image term retained substantially stronger discriminative ranking information than the direct radiomics term. The same structural-to-effective discrepancy was observed in exploratory external evaluation. Branch-targeted permutation, integrated-gradient, and Shapley-style analyses further identified radiomics descriptors influencing the direct radiomics term. Because radiomics also generates the gate, the gated image term remains radiomics-dependent, so additive branch contribution should not be interpreted as total input influence. These findings show that structural gate weighting and decision-level branch contribution represent distinct aspects of gated feature fusion and should be interpreted separately.

eess.IV↗

arXiv2026

Complementary Roles of Radiomics and Foundation Representations in Renal Cell Carcinoma Classification: A Comparative Study of 2D and 3D CT Encodings

Accurate preoperative subtype classification of renal cell carcinoma (RCC) from contrast-enhanced computed tomography remains clinically challenging. Radiomics provides structured tumour descriptors, whereas foundation representations offer transferable image features. However, it remains unclear whether radiomics still adds value beyond pretrained representations, and how 2D and 3D MedVAE encoders compare in this setting. We compared handcrafted radiomics, 2D MedVAE, 3D MedVAE, and their fusion for binary clear-cell RCC versus non-clear-cell RCC classification on KiTS23 under a unified preprocessing pipeline. Concatenation, cross-attention, and gated fusion were evaluated as representative integration strategies, and radiomics feature importance was analysed to support decision-centric interpretability. Fusion consistently improved discrimination over image-only MedVAE branches. The best overall performance was achieved by 3D gated fusion, with an AUC of 82.7\%, outperforming the best 2D fusion model (79.6%), the radiomics baseline (74.4%), and the single-modality MedVAE branches. Ablation analysis further showed clear gains of the full fusion model over both image-only and radiomics-only variants, indicating complementary contributions from radiomics and image representations. These findings suggest that radiomics remains relevant for RCC CT classification in the presence of foundation representations, and that its integration with MedVAE is more effective in the 3D setting. More broadly, the study supports a complementary role for radiomics and foundation representations in clinically meaningful imaging decision support.

cs.CV↗

arXiv2026

Radiomics-Conditioned Modulation of RenalCLIP Features for Clear Cell Renal Cell Carcinoma Classification

Radiomics provides quantitative descriptions of tumour appearance that may complement disease-specific foundation models in small labelled cohorts. We investigate this complementarity for computed tomography-based classification of clear cell renal cell carcinoma. Our framework uses radiomics to modulate RenalCLIP features through feature-wise linear modulation (FiLM), while retaining a direct radiomics contribution. Internal testing and external validation compare it with conventional fusion strategies and reference classifiers. The FiLM model achieves an area under the receiver operating characteristic curve (AUC) of 0.804 internally and 0.854 externally, with the highest mean AUC among the evaluated RenalCLIP fusion strategies in both cohorts. Pathway ablations examine the contributions of conditional modulation and the direct radiomics residual, while feature permutation highlights the role of tumour texture. These findings support radiomics as a useful complement to RenalCLIP in a small labelled cohort and identify FiLM as an effective approach to integrating their representations for robust renal tumour classification.

cs.CV↗

arXiv2026

Radiomics--Foundation Fusion for Interpretable RCC Classification: Internal Benchmarking and Exploratory External Transfer

Accurate preoperative subtype classification of renal cell carcinoma (RCC) from contrast-enhanced CT remains clinically challenging because clear cell RCC (ccRCC) and non-clear cell RCC often show overlapping imaging appearances. This study evaluates whether foundation representations reduce reliance on handcrafted radiomics, or whether radiomics remains complementary for interpretable tumour characterisation. We compared radiomics, conventional CNN features, MedicalNet-pretrained features, MedVAE representations, and fusion variants for binary ccRCC classification on KiTS23, reporting area under the receiver operating characteristic curve (AUC) with bootstrap confidence intervals and average precision (AP) as a complementary class-imbalance-sensitive metric. We further assessed branch-removal ablation, TCGA/AIMI external transfer, and interpretability using radiomics permutation importance and gate-level analysis. Internally, 3D MedVAE gated fusion achieved the best performance, with an AUC of 82.7% and AP of 92.2%. On the external TCGA cohort, the same model achieved an AUC of 79.5% and AP of 98.9%, although specificity remains uncertain because only two external non-ccRCC cases were available. Gate analysis showed a radiomics-dominant fusion regime, suggesting that foundation representations acted as case-dependent refinement signals rather than replacements for structured tumour descriptors. These findings support radiomics as a complementary and clinically interpretable component of CT-based RCC characterisation in the foundation-model era.

cs.CV↗

arXiv2026

A Hybrid Anyon-Otto thermal machine

We propose a four-stroke quantum thermal machine based on the 1D anyon Hubbard model, which is capable of extracting the excess energy arising from anyon exclusion statistics at low temperature into finite work. Defining a hybrid anyon-Otto (HAO) cycle, we find that the low-temperature work, in the absence of any interactions, is maximized in the pseudo-fermionic limit, where the anyons most closely resemble free fermions. However, when weak interactions are introduced, the work output is no longer maximized at the bosonic or pseudo-fermionic extremes but instead peaks at intermediate statistical angles. This clearly demonstrates that interactions and anyonic statistics conspire non-trivially to enhance performance, with interacting anyons offering greater quantum thermodynamic advantage than either bosons or pseudo-fermions, in this regime. Furthermore, we also outline an experimental protocol to realize the HAO cycle using ultracold atoms in an optical lattice.

cond-mat.str-el↗

arXiv2025

Fusing Radiomic Features with Deep Representations for Gestational Age Estimation in Fetal Ultrasound Images

Accurate gestational age (GA) estimation, ideally through fetal ultrasound measurement, is a crucial aspect of providing excellent antenatal care. However, deriving GA from manual fetal biometric measurements depends on the operator and is time-consuming. Hence, automatic computer-assisted methods are demanded in clinical practice. In this paper, we present a novel feature fusion framework to estimate GA using fetal ultrasound images without any measurement information. We adopt a deep learning model to extract deep representations from ultrasound images. We extract radiomic features to reveal patterns and characteristics of fetal brain growth. To harness the interpretability of radiomics in medical imaging analysis, we estimate GA by fusing radiomic features and deep representations. Our framework estimates GA with a mean absolute error of 8.0 days across three trimesters, outperforming current machine learning-based methods at these gestational ages. Experimental results demonstrate the robustness of our framework across different populations in diverse geographical regions. Our code is publicly available on \href{https://github.com/13204942/RadiomicsImageFusion_FetalUS}.

eess.IV↗

arXiv2025

Anderson localization induced by structural disorder

We examine the onset of Anderson localization in three-dimensional systems with structural disorder in the form of lattice irregularities and in the absence of any on-site disordered potential. Analyzing two models with distinct types of lattice regularities, we show that the Anderson localization transition occurs when the strength of the structural disorder is smoothly increased. Performing finite-size scaling analysis of the results, we show that the transition belongs to the same universality class as regular Anderson localization induced by onsite disorder. Our work identifies a new class of structurally disordered lattice models in which destructive interference of matter waves may inhibit transport and lead to a transition between metallic and localized phases.

cond-mat.dis-nn↗

arXiv2024

Sharp detection of the onset of Floquet heating using eigenstate sensitivity

Chaotic Floquet systems at sufficiently low driving frequencies are known to heat up to an infinite temperature ensemble in the thermodynamic limit. However at high driving frequencies, Floquet systems remain energetically stable in a robust prethermal phase with exponentially long heating times. We propose sensitivity (susceptibility) of Floquet eigenstates against infinitesimal deformations of the drive, as a sharp and sensitive measure to detect this heating transition. It also captures various regimes (timescales) of Floquet thermalization accurately. Particularly, we find that at low frequencies near the onset of unbounded heating, Floquet eigenstates are maximally sensitive to perturbations and consequently the scaled susceptibility develops a sharp maximum. We further connect our results to the relaxation dynamics of local observables to show that near the onset of Floquet heating, the system is nonergodic with slow glassy dynamics despite being nonintegrable at all driving frequencies.

cond-mat.stat-mech↗

arXiv2024

Signatures of quantum phases in a dissipative system

Lindbladian formalism, as tuned to dissipative and open systems, has been all-pervasive to interpret non-equilibrium steady states of quantum many-body systems. We study the fate of free fermionic and superconducting phases in a dissipative one-dimensional Kitaev model - where the bath acts both as a source and a sink of fermionic particles with different coupling rates. As a function of these two couplings, we investigate the steady state, its entanglement content, and its approach from varying initial states. Interestingly, we find that the steady state phase diagram retains decipherable signatures of ground state critical physics. We also show that early-time fidelity is a useful marker to find a subclass of phase transitions in such situations. Moreover, we show that the survival of critical signatures at late-times, strongly depend on the thermal nature of the steady state. This connection hints at a correspondence between quantum observables and classical magnetism in the steady state of such systems. Our work uncovers interesting connections between dissipative quantum many-body systems, thermalization of a classical spin and many-body quantum critical phenomena.

cond-mat.str-el↗

arXiv2024

Slow relaxation of quasi-periodically driven integrable quantum many-body systems

We study the emergence and stability of a prethermal phase in an integrable many-body system subjected to a Fibonacci drive. Despite not being periodic, Fibonacci drives have been shown to introduce dynamical constraints due to their self-similar structure, unlike random driving protocols. From perturbative analysis, this has been argued to result in an exponentially long prethermal phase in the high frequency limit of driving. Examining higher order terms in the perturbative expansion, we show that the perturbative description breaks down eventually in such systems at a finite universal order, which depends solely on features of the Fibonacci sequence. This leads to an onset of energy absorption at long time scales for intermediate and low driving frequencies. Interestingly, in spite of the breakdown of an effective Hamiltonian in the perturbative analysis, we still observe slow logarithmic heating time-scales, unlike purely random drives.

cond-mat.stat-mech↗

arXiv2022

Quasi-localization dynamics in a Fibonacci quantum rotor

We analyze the dynamics of a quantum kicked rotor (QKR) driven with a binary Fibonacci sequence of two distinct drive amplitudes. While the dynamics at low drive frequencies is found to be diffusive, a long-lived pre-ergodic regime emerges in the other limit. Further, the dynamics in this pre-ergodic regime can be associated with the onset of a dynamical quasi-localization, similar to the dynamical localization observed in a regular QKR. We establish that this peculiar behavior arises due to the presence of localized eigenstates of an approximately conserved effective Hamiltonian, which drives the evolution at Fibonacci instants. However, the effective Hamiltonian picture does not persist indefinitely and the dynamics eventually becomes ergodic after asymptotically long times.

quant-ph↗

arXiv2022

Periodically driven many-body quantum battery

We explore the charging of a quantum battery based on spin systems through periodic modulation of a transverse-field like Ising Hamiltonian. In the integrable limit, we find that resonance tunneling can lead to a higher transfer of energy to the battery and better stability of the stored energy at specific drive frequencies. When the integrability is broken in the presence of an additional longitudinal field, we find that the effective Floquet Hamiltonian contains terms which may lead to a global charging of the battery. However, we do not find any quantum advantage in the charging power, thus demonstrating that global charging is only a necessary and not sufficient condition for achieving quantum advantage.

quant-ph↗

arXiv2022

Detecting topological phase transitions through entanglement between disconnected partitions in a Kitaev chain with long-range interactions

We explore the behaviour of the disconnected entanglement entropy (DEE) across the topological phases of a long range interacting Kitaev chain where the long range interactions decay as a power law with an exponent $α$. We show that while the DEE may not remain invariant deep within the topologically non-trivial phase when $α<1$, it nevertheless shows a quantized discontinuous jump at the quantum critical point and can act as a strong marker for the detection of topological phase transition. We also study the time evolution of the DEE after a sudden quench of the chemical potential within the same phase. In the short range limit of a finite chain, the DEE is expected to remain constant upto a critical time after the quench, which diverges in the thermodynamic limit. However, no such critical time is found to exist when the long range interactions dominate (i.e., $α<1$).

cond-mat.stat-mech↗

arXiv2021

Quantum thermal machines and batteries

The seminal work by Sadi Carnot in the early nineteenth century provided the blueprint of a reversible heat engine and the celebrated second law of thermodynamics eventually followed. Almost two centuries later, the quest to formulate a quantum theory of the thermodynamic laws has thus unsurprisingly motivated physicists to visualise what are known as `quantum thermal machines' (QTMs). In this article, we review the prominent developments achieved in the theoretical construction as well as understanding of QTMs, beginning from the formulation of their earliest prototypes to recent models. We also present a detailed introduction and highlight recent progress in the rapidly developing field of `quantum batteries'.

quant-ph↗

arXiv2021

Driven quantum many-body systems and out-of-equilibrium topology

In this review we present some of the work done in India in the area of driven and out-of-equilibrium systems with topological phases. After presenting some well-known examples of topological systems in one and two dimensions, we discuss the effects of periodic driving in some of them. We discuss the unitary as well as the non-unitary dynamical preparation of topologically non-trivial states in one and two dimensional systems. We then discuss the effects of Majorana end modes on transport through a Kitaev chain and a junction of three Kitaev chains. Transport through the surface states of a three-dimensional topological insulator is discussed. The effects of hybridization between the top and bottom surfaces and the application of electromagnetic radiation on a strip-like region on the top surface are described. Two unusual topological systems are mentioned briefly, namely, a spin system on a kagome lattice and a Josephson junction of three superconducting wires. We have also included a pedagogical discussion on topology and topological invariants in the appendices, where the connection between topological properties and the intrinsic geometry of quantum states is also elucidated.

cond-mat.stat-mech↗

arXiv2021

Bilayer Haldane system: Topological characterization and adiabatic passages connecting Chern phases

We present a complete topological characterization of a bilayer composite of two Chern insulators (specifically, Haldane models) and explicitly establish the bulk-boundary correspondences. We show that an appropriately defined Chern number accurately maps out all the possible phases of the system and remains well-defined even in the presence of degeneracies in the occupied bands. Importantly, our result paves the way for realizing adiabatic preparation of monolayer Chern insulators. This has been a major challenge till date, given the impossibility of unitarily connecting inequivalent topological phases. We show that this difficulty can be circumvented by adiabatically varying the interlayer coupling in such a way that the system remains gapped at all times. In particular, a complete knowledge of the phase diagram of the bilayer composite immediately allows one to identify all such adiabatic passages which may connect the different Chern inequivalent phases of the individual monolayers.

cond-mat.stat-mech↗

arXiv2020

Dynamical generation of Majorana edge correlations in a ramped Kitaev chain coupled to nonthermal dissipative channels

We quantitatively study the out-of-equilibrium edge-Majorana correlation in a linearly ramped one-dimensional Kitaev chain of finite length in a dissipative environment. The chemical potential is dynamically ramped to drive the chain from its topologically trivial to nontrivial phase in the presence of couplings to nonthermal Markovian baths. We consider two distinctive situations: In the first situation, the bath is quasilocal in the site basis (local in quasiparticle basis) while in the other it is local. Following a Lindbladian approach, we compute the early time dynamics as well as the asymptotic behavior of the edge-Majorana correlation to probe the interplay between two competing timescales - one due to the coherent ramping while the other to the dissipative coupling. For the quasilocal bath, we establish that there is a steady generation of Majorana correlations in asymptotic time and the presence of an optimal ramping time which facilitates a quicker approach to the topological steady state. In the second scenario, we analyze the action of a local particle-loss type of bath in which we have established the existence of an optimal ramping time which results from the competing dynamics between the unitary ramp and the dissipative coupling. While the defect generated by the former decays exponentially with increasing ramp duration, the later scales linearly with the same. This linear scaling is further established through a perturbation theory formulated using the nondimensionalized coupling to the bath as a small parameter.

cond-mat.stat-mech↗