Regression Concept Vectors for Bidirectional Explanations in Histopathology

By Mara Graziani, Vincent Andrearczyk, Henning Müller. 2018

89 citations recorded by the source corpus.

Explanations for deep neural network predictions in terms of domain-related concepts can be valuable in medical applications, where justifications are important for confidence in the decision-making. In this work, we propose a methodology to exploit continuous concept measures as Regression Concept Vectors (RCVs) in the activation space of a layer. The directional derivative of the decision function along the RCVs represents the network sensitivity to increasing values of a given concept measure. When applied to breast cancer grading, nuclei texture emerges as a relevant concept in the detection of tumor tissue in breast lymph node samples. We evaluate score robustness and consistency by statistical analysis.

Subfields: Artificial Intelligence, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Percy Liang
2. Cuntai Guan
3. Erico Tjoa
4. Been Kim
5. Pang Wei Koh
6. Mohit Bansal
7. Noah A. Smith
8. Yejin Choi

In Radiology, Nuclear Medicine and Imaging

1. Amitojdeep Singh
2. Percy Liang
3. Cuntai Guan
4. Pang Wei Koh
5. Been Kim
6. Vasudevan Lakshminarayanan
7. Erico Tjoa
8. Sourya Sengupta

Citations in the corpus

• A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI (2019, methodology): “brain, neuro* breast [64], [69], [139], [140] [65], [66], [82], [90], [99] skin [138], heart [123] [113], [122], [134], [156] others [43], [66], [137], [141] Comp.”
• Concept Bottleneck Models (2020, positive): “A flurry of recent papers have used similar human concepts for post-hoc interpretation of medical and other scientific ML models, e.g., Graziani et al. (2018) for breast cancer histopathology; Clough et al. (2019) for cardiac MRIs; and Sprague et al. (2019) for meteorology (storm prediction).”
• Acquisition of chess knowledge in AlphaZero (2021, background): “Concept-based explanations have been successfully used in complex scientific and medical domains, as seen in [20, 21, 22, 23, 24, 25, 26, 27].”
• Explainable Deep Learning Models in Medical Image Analysis (2020, methodology): “An extension of TCAV, which used the presence or absence of concepts, using Regression Concept Vectors (RCV) in the activation space of a layer was used to detect continuous concepts [57].”