TTE-CAM: Self-Explainable Class Activation Maps for Pretrained Black-Box CNNs

Kerol Djoumessi; Philipp Berens

cs.CVMar 27, 2026

TTE-CAM: Self-Explainable Class Activation Maps for Pretrained Black-Box CNNs

Authors: Kerol Djoumessi, Philipp Berens

Organizations: Hertie Institute for AI in Brain Health, University of Tübingen, Germany · Tübingen AI Center, University of Tübingen, Germany

Abstract

Convolutional neural networks (CNNs) achieve state-of-the-art performance in medical image analysis yet remain opaque, limiting adoption in high-stakes clinical settings. Existing approaches face a fundamental trade-off: post-hoc methods provide unfaithful approximate explanations, while inherently interpretable architectures are faithful but often sacrifice predictive performance. We introduce TTE-CAM, a test-time framework that bridges this gap by converting pretrained black-box CNNs into self-explainable models via a convolution-based replacement of their classification head, initialized from the original weights. The resulting model preserves black-box predictive performance while delivering built-in faithful explanations competitive with post-hoc methods, both qualitatively and quantitatively. The code is available at https://github.com/kdjoumessi/Test-Time-Explainability

Figures & tables

Figure 1: Qualitative explanations comparison. The first column shows a DR fundus with clinical annotations (green markers) and a pneumonia CXR with ground-truth bounding boxes (green). Columns 2-6 show post-hoc saliency maps; the last column shows TTE-CAM explanations. CXR scores indicate activation precision. Both samples were randomly selected from their corresponding disease classes.

	Metrics	CAM	S. CAM	L. CAM	G. CAM	IG	TTE-CAM
Fundus	Topk Prec. $\uparrow$	$.33\pm.29$	$.28\pm.22$	$.31\pm.26$	$.39\pm.28$	$.39\pm.28$	$.33\pm.28$
Fundus	Topk Sens. $\downarrow$	$0.629$	$0.668$	$0.644$	$0.629$	$0.605$	$0.629$
CXR	Acti. Prec. $\uparrow$	$.12\pm.09$	$.13\pm.10$	$.12\pm.09$	$.13\pm.10$	$.12\pm.09$	$.12\pm.09$
CXR	Topk Sens. $\downarrow$	$0.953$	$0.959$	$0.955$	$0.953$	$0.963$	$0.953$

Table 1: Quantitative explanation comparison.

\uparrow

higher is better;

\downarrow

lower is better.

Explore similar work

Med-CAM: Minimal Evidence for Explaining Medical Decision Making
Apr 15, 2026Pirzada Suhail, Aditya Anand, Amit SethiExplainable Image Classification Explainable Medical Image Analysis
IIT Bombay
Post-hoc Self-explanation of CNNs
Mar 30, 2026Ahcène Boubekki, Line H. ClemmensenNeural Network Interpretability Explainable Image Classification
Department of Mathematical Sciences, University of Copenhagen, Denmark
Class Activation Mapping in Explainable Computer Vision: A Method-Centered Review of CNN, Transformer, and Foundation-Model-Era Visual Explanations
Aug 12, 2026AmirHossein Eshghi, Hamid Saadatfar, Seyyed Ali Hoseini +2Transformer Interpretability Feature Attribution
Department of Computer Engineering, University of Birjand, Iran

Apr 15, 2026J/K move · Enter open · S save

Med-CAM: Minimal Evidence for Explaining Medical Decision Making

Pirzada Suhail, Aditya Anand, Amit Sethi

IIT Bombay

Explainable Image Classification Explainable Medical Image Analysis Feature Attribution

Open paper AskPDF arXiv

Reliable and interpretable decision-making is essential in medical imaging, where diagnostic outcomes directly influence patient care. Despite advances in deep learning, most medical AI systems operate as opaque black boxes, providing little insight into why a particular diagnosis was reached. In this paper, we introduce Med-CAM, a framework for generating minimal and sharp maps as evidence-based explanations for Medical decision making via Classifier Activation Matching. Med-CAM trains a segmentation network from scratch to produce a mask that highlights the minimal evidence critical to model's decision for any seen or unseen image. This ensures that the explanation is both faithful to the network's behaviour and interpretable to clinicians. Experiments show, unlike prior spatial explanation methods, such as Grad-CAM and attention maps, which yield only fuzzy regions of relative importance, Med-CAM with its superior spatial awareness to shapes, textures, and boundaries, delivers conclusive, evidence-based explanations that faithfully replicate the model's prediction for any given image. By explicitly constraining explanations to be compact, consistent with model activations, and diagnostic alignment, Med-CAM advances transparent AI to foster clinician understanding and trust in high-stakes medical applications such as pathology and radiology.

Abstract

Explore similar work

Med-CAM: Minimal Evidence for Explaining Medical Decision Making

Post-hoc Self-explanation of CNNs

Class Activation Mapping in Explainable Computer Vision: A Method-Centered Review of CNN, Transformer, and Foundation-Model-Era Visual Explanations