pith. sign in

arxiv: 2603.00223 · v2 · submitted 2026-02-27 · 💻 cs.CV · quant-ph

Pretty Good Measurement for Radiomics: A Quantum-Inspired Multi-Class Classifier for Lung Cancer Subtyping and Prostate Cancer Risk Stratification

Giuseppe Sergioli , Carlo Cuccu , Giovanni Pasini , Alessandro Stefano , Giorgio Russo , Andr\'es Camilo Granda Arango , Roberto Giuntini This is my paper

Pith reviewed 2026-05-15 17:58 UTC · model grok-4.3

classification 💻 cs.CV quant-ph
keywords Pretty Good Measurementquantum-inspired classificationradiomicsNSCLC subtypingprostate cancer stratificationmulti-class classifierPOVM
0
0 comments X

The pith

A quantum-inspired classifier encodes classes as mixed states and uses a single Pretty Good Measurement to handle multi-class radiomics tasks directly.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents a supervised multi-class classifier derived from quantum state discrimination, where each class is represented by an encoded mixed state and decisions are made via one POVM constructed from the Pretty Good Measurement. This replaces the usual reduction to binary subproblems with a geometry-driven rule that depends on how the encoding map positions the class states and on their mutual overlaps. When applied to radiomics feature vectors from non-small-cell lung carcinoma subtyping and prostate cancer risk stratification, the resulting classifier matches or exceeds standard classical baselines in accuracy and in clinically relevant sensitivity-specificity balance, with clearest gains on the NSCLC binary and three-class problems. A reader should care because the method supplies an alternative decision geometry that may prove more stable when feature spaces contain substantial class overlap.

Core claim

Classification is recast as discrimination among a finite ensemble of class-dependent density operators; the Pretty Good Measurement supplies the single POVM that realizes the decision rule, and empirical performance on the two radiomics datasets is governed by the overlap structure induced by the chosen encoding map.

What carries the argument

The Pretty Good Measurement viewed as an operator-valued decision rule that constructs one POVM for the entire ensemble of class-encoded mixed states.

If this is right

  • The approach supplies a genuinely multi-class procedure that never reduces to pairwise or one-versus-rest subproblems.
  • Classification accuracy is controlled by the geometry of the encoding map and the overlap pattern among the class states.
  • On the NSCLC tasks the method improves upon several standard baselines; on the four-class and PCa tasks it remains competitive while preserving useful sensitivity-specificity trade-offs.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same encoding-plus-POVM construction could be tested on other high-dimensional medical imaging tasks where class boundaries are known to be soft.
  • Feature-selection routines might be redesigned to minimize the overlap measure that the PGM explicitly uses.
  • Hybrid pipelines that feed classical radiomics features into a PGM stage could be benchmarked against end-to-end deep networks on the same datasets.

Load-bearing premise

Encoding classes as mixed states and deriving one POVM from quantum state discrimination will capture enough of the relevant overlap geometry to produce reliable gains over classical methods in radiomics feature spaces.

What would settle it

On the same NSCLC three-class dataset and feature-selection protocol, if the PGM classifier records accuracy or macro-F1 scores materially below the strongest classical ensemble baseline, the claim of consistent competitiveness would be refuted.

Figures

Figures reproduced from arXiv: 2603.00223 by Alessandro Stefano, Andr\'es Camilo Granda Arango, Carlo Cuccu, Giorgio Russo, Giovanni Pasini, Giuseppe Sergioli, Roberto Giuntini.

Figure 1
Figure 1. Figure 1: In the first row (NSCLC Case Study) there is an example of CT at the three anatomical [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Comparative evaluation of the proposed PGM classifier against conventional baseline [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparative evaluation of the proposed PGM classifier against conventional baseline [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparative evaluation of the proposed PGM classifier against conventional baseline [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Win–loss heatmap comparing classifiers for each labeling scenario. The three labeling [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Performance analysis of the PGM classifier on the PCa dataset under different feature [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Heat-map of metric differences (PGM − Ensemble) across feature-selection scenarios. Each row corresponds to a feature-set, and each column represents one of the six evaluation metrics Accuracy, AUC, Sensitivity, Specificity, Precision, and F-score. Cells are coloured on a diverging blue–red scale centred at zero: A blue hue indicates that the PGM classifier outperforms the ensemble for that metric (positiv… view at source ↗
read the original abstract

We investigate a quantum-inspired approach to supervised multi-class classification based on the Pretty Good Measurement (PGM), viewed as an operator-valued decision rule derived from quantum state discrimination. The method associates each class with an encoded mixed state and performs classification through a single POVM construction, thus providing a genuinely multi-class strategy without reduction to pairwise or one-vs-rest schemes. In this perspective, classification is reformulated as the discrimination of a finite ensemble of class-dependent density operators, with performance governed by the geometry induced by the encoding map and by the overlap structure among classes. To assess the practical scope of this framework, we apply the PGM-based classifier to two biomedical radiomics case studies: histopathological subtyping of non-small-cell lung carcinoma (NSCLC) and prostate cancer (PCa) risk stratification. The evaluation is conducted under protocols aligned with previously reported radiomics studies, enabling direct comparison with established classical baselines. The results show that the PGM-based classifier is consistently competitive and, in several settings, improves upon standard methods. In particular, the method performs especially well in the NSCLC binary and three-class tasks, while remaining competitive in the four-class case, where increased class overlap yields a more demanding discrimination geometry. In the PCa study, the PGM classifier remains close to the strongest ensemble baseline and exhibits clinically relevant sensitivity--specificity trade-offs across feature-selection scenarios.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 3 minor

Summary. The manuscript introduces a quantum-inspired multi-class classifier based on the Pretty Good Measurement (PGM) derived from quantum state discrimination. Each class is encoded as a mixed quantum state via a feature-to-density-operator map, after which classification is performed by a single POVM constructed from the ensemble of states. The method is evaluated on two radiomics tasks—NSCLC histopathological subtyping (binary and multi-class) and prostate cancer risk stratification—under protocols aligned with prior studies, with results indicating competitive or superior performance relative to standard classical baselines, particularly in the NSCLC settings.

Significance. If the reported performance holds under rigorous statistical scrutiny, the work supplies a genuinely multi-class quantum-inspired decision rule that avoids pairwise or one-vs-rest reductions and directly exploits class-overlap geometry. This could be valuable in radiomics domains where class-conditional distributions exhibit substantial overlap. The explicit alignment with established protocols and the parameter-free character of the PGM construction (once the encoding is fixed) are notable strengths that facilitate reproducibility and direct benchmarking.

major comments (2)
  1. [§4.2] §4.2 (NSCLC three-class results): the manuscript reports that PGM outperforms standard methods, yet the quantitative table provides only point estimates without cross-validation standard deviations, confidence intervals, or paired statistical tests (e.g., McNemar or Wilcoxon). This weakens the claim that the method “improves upon” baselines, as the observed margins could be within experimental variability.
  2. [§3.3] §3.3 (encoding map): the mixed-state construction is defined via a kernel or feature map whose hyperparameters (e.g., scaling or regularization) are not explicitly declared as fixed or cross-validated. If any such choice is data-dependent, the “parameter-free” character of the subsequent PGM step is compromised and must be clarified to support the circularity assessment.
minor comments (3)
  1. [Figure 2] Figure 2 (POVM visualization): the color scale and axis labels are insufficiently described; add a caption clarifying what the plotted operators represent (e.g., diagonal elements of the measurement operators).
  2. [Table 1] Table 1 (PCa cohort): the feature-selection scenarios are listed but the exact number of retained features per scenario is omitted; include these counts for reproducibility.
  3. [§5] §5 (discussion): the clinical relevance of the sensitivity–specificity trade-offs is asserted but not quantified against established diagnostic thresholds; a brief comparison to literature-reported operating points would strengthen the interpretation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and positive overall assessment. We address each major point below and have revised the manuscript accordingly to strengthen statistical reporting and clarify the encoding procedure.

read point-by-point responses

  1. Referee: [§4.2] §4.2 (NSCLC three-class results): the manuscript reports that PGM outperforms standard methods, yet the quantitative table provides only point estimates without cross-validation standard deviations, confidence intervals, or paired statistical tests (e.g., McNemar or Wilcoxon). This weakens the claim that the method “improves upon” baselines, as the observed margins could be within experimental variability.

    Authors: We agree that point estimates alone limit the strength of the performance claims. In the revised manuscript we have recomputed all NSCLC results under 5-fold cross-validation, now reporting mean accuracy together with standard deviations and 95% confidence intervals. We have also added McNemar tests for paired comparisons against each baseline, with p-values reported in the updated Table 2. These additions confirm that the observed margins in the three-class setting remain statistically significant at the 0.05 level. revision: yes

  2. Referee: [§3.3] §3.3 (encoding map): the mixed-state construction is defined via a kernel or feature map whose hyperparameters (e.g., scaling or regularization) are not explicitly declared as fixed or cross-validated. If any such choice is data-dependent, the “parameter-free” character of the subsequent PGM step is compromised and must be clarified to support the circularity assessment.

    Authors: The encoding map uses a fixed Gaussian kernel whose single scaling parameter is set deterministically from the input feature dimension (σ = √d) and is never tuned on the target dataset. No regularization or other data-dependent choices are involved. We have inserted an explicit paragraph in §3.3 stating these choices and confirming that the PGM construction itself remains parameter-free once the encoding is fixed, thereby preserving the non-circular evaluation protocol used throughout the experiments. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation follows established quantum state discrimination

full rationale

The paper constructs the PGM classifier directly from standard quantum information theory by associating each class with an encoded mixed state and deriving a single POVM for discrimination. No equation or protocol reduces by construction to a fitted parameter defined from the target radiomics data, nor does any central claim rest on a self-citation chain that itself lacks independent verification. Performance claims are supported by empirical comparisons under aligned radiomics protocols rather than tautological redefinitions. The encoding geometry and overlap structure are treated as inputs to the discrimination task, not outputs derived from the classification results themselves.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard quantum information assumptions about state encoding and POVM construction; no free parameters or invented entities are introduced in the abstract description.

axioms (2)
  • domain assumption Each class can be represented by an encoded mixed state such that discrimination geometry reflects class overlap
    Invoked when associating classes with density operators and deriving the POVM decision rule.
  • domain assumption A single POVM suffices for multi-class discrimination without reduction to pairwise schemes
    Core to the reformulation of classification as quantum state discrimination.

pith-pipeline@v0.9.0 · 5581 in / 1161 out tokens · 51543 ms · 2026-05-15T17:58:57.113456+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    classification is reformulated as the discrimination of a finite ensemble of class-dependent density operators, with performance governed by the geometry induced by the encoding map and by the overlap structure among classes

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

  1. [1]

    doi:10.3389/fninf.2016.00017

    Diederik Aerts, Massimiliano Sassoli de Bianchi, and Sandro Sozzo. “On the Foundations of the Brussels Operational-Realistic Approach to Cognition”. In:Frontiers in Physics4 (May 2016).issn: 2296-424X.doi:10.3389/fphy.2016.00017

    work page doi:10.3389/fphy.2016.00017 2016

  2. [2]

    Quantum structure in cognition

    Diederik Aerts. “Quantum structure in cognition”. In:Journal of Mathematical Psychology 53.5 (2009). Special Issue: Quantum Cognition, pp. 314–348.issn: 0022-2496.doi:10.1016/ j.jmp.2009.04.005

    work page 2009

  3. [3]

    The Algebraic Structure of an Approximately Universal System of Quantum Computational Gates

    M.L. Dalla Chiara et al. “The Algebraic Structure of an Approximately Universal System of Quantum Computational Gates”. In:Foundations of Physics39.6 (2009), pp. 559–572.issn: 1572-9516.doi:10.1007/s10701-009-9302-0

    work page doi:10.1007/s10701-009-9302-0 2009

  4. [4]

    Quantum Cognition Beyond Hilbert Space: Fundamentals and Applica- tions

    Diederik Aerts et al. “Quantum Cognition Beyond Hilbert Space: Fundamentals and Applica- tions”. In:Quantum Interaction. Springer International Publishing, 2017, pp. 81–98.isbn: 9783319522890.doi:10.1007/978-3-319-52289-0_7

    work page doi:10.1007/978-3-319-52289-0_7 2017

  5. [5]

    Modeling Human Decision-Making: An Overview of the Brussels Quantum Approach

    Diederik Aerts et al. “Modeling Human Decision-Making: An Overview of the Brussels Quantum Approach”. In:Foundations of Science26.1 (Mar. 2021), pp. 27–54.issn: 1572-8471. doi:10.1007/s10699-018-9559-x

    work page doi:10.1007/s10699-018-9559-x 2021

  6. [6]

    From quantum cogni- tion to conceptuality interpretation I: tracing the Brussels group’s intellectual journey

    Diederik Aerts, Massimiliano Sassoli de Bianchi, and Sandro Sozzo. “From quantum cogni- tion to conceptuality interpretation I: tracing the Brussels group’s intellectual journey”. In: Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences383.2309 (Nov. 2025).issn: 1471-2962.doi:10.1098/rsta.2024.0382

    work page doi:10.1098/rsta.2024.0382 2025

  7. [7]

    Testing Quantum Models of Conjunction Fallacy on the World Wide Web

    Diederik Aerts et al. “Testing Quantum Models of Conjunction Fallacy on the World Wide Web”. In:International Journal of Theoretical Physics56.12 (Dec. 2017), pp. 3744–3756.issn: 1572-9575.doi:10.1007/s10773-017-3288-8

    work page doi:10.1007/s10773-017-3288-8 2017

  8. [8]

    A many-valued approach to quantum computational logics

    M.L. Dalla Chiara et al. “A many-valued approach to quantum computational logics”. In: Fuzzy Sets and Systems335 (2018). Special Issue: Selected Papers from the 36th Linz Seminar on Fuzzy Set Theory, pp. 94–111.issn: 0165-0114.doi:10.1016/j.fss.2016.12.015

    work page doi:10.1016/j.fss.2016.12.015 2018

  9. [9]

    A Quantum-inspired Version of the Classification Problem

    Giuseppe Sergioli et al. “A Quantum-inspired Version of the Classification Problem”. In: International Journal of Theoretical Physics56.12 (Dec. 2017), pp. 3880–3888.issn: 1572- 9575.doi:10.1007/s10773-017-3371-1

    work page doi:10.1007/s10773-017-3371-1 2017

  10. [10]

    Caziot and B

    Giuseppe Sergioli, Roberto Giuntini, and Hector Freytes. “A new quantum approach to binary classification”. In:PLOS ONE14.5 (May 2019), pp. 1–14.doi:10.1371/journal.pone. 0216224

    work page doi:10.1371/journal.pone 2019

  11. [11]

    A quantum-inspired classifier for clonogenic assay evaluations

    Giuseppe Sergioli et al. “A quantum-inspired classifier for clonogenic assay evaluations”. In: Scientific Reports11.1 (2021), p. 2830.issn: 2045-2322.doi: 10.1038/s41598-021-82085-8

    work page doi:10.1038/s41598-021-82085-8 2021

  12. [12]

    Quantum-inspired algorithm for direct multi-class classification,

    Roberto Giuntini et al. “Quantum-inspired algorithm for direct multi-class classification”. In: Applied Soft Computing134 (2023), p. 109956.doi:10.1016/j.asoc.2022.109956

    work page doi:10.1016/j.asoc.2022.109956 2023

  13. [13]

    Multi-class classification based on quantum state discrimination

    Roberto Giuntini et al. “Multi-class classification based on quantum state discrimination”. In: Fuzzy Sets and Systems467 (2023), p. 108509.issn: 0165-0114.doi: 10.1016/j.fss.2023. 03.012

    work page doi:10.1016/j.fss.2023 2023

  14. [14]

    Phenotyping the Histopathological Subtypes of Non-Small-Cell Lung Carcinoma: How Beneficial Is Radiomics?

    Giovanni Pasini et al. “Phenotyping the Histopathological Subtypes of Non-Small-Cell Lung Carcinoma: How Beneficial Is Radiomics?” In:Diagnostics13.6 (2023).issn: 2075-4418.doi: 10.3390/diagnostics13061167

    work page doi:10.3390/diagnostics13061167 2023

  15. [15]

    A Robust [18F]-PSMA-1007 Radiomics Ensemble Model for Prostate Cancer Risk Stratification

    Giovanni Pasini et al. “A Robust [18F]-PSMA-1007 Radiomics Ensemble Model for Prostate Cancer Risk Stratification”. In:Journal of Imaging Informatics in Medicine38.3 (2025), pp. 1388–1402.issn: 2948-2933.doi:10.1007/s10278-024-01281-w

    work page doi:10.1007/s10278-024-01281-w 2025

  16. [16]

    Radiomics: Extracting more information from medical images using advanced feature analysis

    Philippe Lambin et al. “Radiomics: Extracting more information from medical images using advanced feature analysis”. In:European Journal of Cancer48.4 (2012), pp. 441–446.issn: 0959-8049.doi:10.1016/j.ejca.2011.11.036

    work page doi:10.1016/j.ejca.2011.11.036 2012

  17. [17]

    Challenges and limitations in applying radiomics to PET imaging: Possible opportunities and avenues for research

    Alessandro Stefano. “Challenges and limitations in applying radiomics to PET imaging: Possible opportunities and avenues for research”. In:Computers in Biology and Medicine179 (2024), p. 108827.issn: 0010-4825.doi:10.1016/j.compbiomed.2024.108827

    work page doi:10.1016/j.compbiomed.2024.108827 2024

  18. [18]

    van der Laak, Bram van Ginneken, and Clara I

    Geert Litjens et al. “A survey on deep learning in medical image analysis”. In:Medical Image Analysis42 (2017), pp. 60–88.issn: 1361-8415.doi:10.1016/j.media.2017.07.005

    work page doi:10.1016/j.media.2017.07.005 2017

  19. [19]

    doi:10.1038/s41586-019-0980-2

    Vojtěch Havlíček etal. “Supervisedlearning withquantum-enhancedfeature spaces”. In:Nature 567.7747 (Mar. 2019), pp. 209–212.issn: 1476-4687.doi:10.1038/s41586-019-0980-2. 14

    work page doi:10.1038/s41586-019-0980-2 2019

  20. [20]

    Masset, R

    Richard Y Li et al. “Quantum processor-inspired machine learning in the biomedical sciences”. In:Patterns (New York, N.Y.)2.6 (June 2021), p. 100246.issn: 2666-3899.doi: 10.1016/j. patter.2021.100246

    work page doi:10.1016/j 2021

  21. [21]

    Quantum com- puting in bioinformatics: a systematic review mapping

    Katarzyna Nałęcz-Charkiewicz, Kamil Charkiewicz, and Robert M Nowak. “Quantum com- puting in bioinformatics: a systematic review mapping”. In:Briefings in Bioinformatics25.5 (Aug. 2024), bbae391.issn: 1477-4054.doi:10.1093/bib/bbae391

    work page doi:10.1093/bib/bbae391 2024

  22. [22]

    A systematic review of quantum machine learning for digital health

    Riddhi S. Gupta et al. “A systematic review of quantum machine learning for digital health”. In:npj Digital Medicine8.1 (May 2025), p. 237.issn: 2398-6352.doi: 10.1038/s41746-025- 01597-z

    work page doi:10.1038/s41746-025- 2025

  23. [23]

    Non-Small Cell Lung Cancer: Epidemiology, Screening, Diagnosis, and Treatment

    Narjust Duma, Rafael Santana-Davila, and Julian R Molina. “Non-Small Cell Lung Cancer: Epidemiology, Screening, Diagnosis, and Treatment”. In:Mayo Clinic Proceedings94.8 (Aug. 2019), pp. 1623–1640.issn: 0025-6196.doi:10.1016/j.mayocp.2019.01.013

    work page doi:10.1016/j.mayocp.2019.01.013 2019

  24. [24]

    The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping

    Alex Zwanenburg et al. “The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping”. In:Radiology295.2 (2020). PMID: 32154773, pp. 328–338.doi:10.1148/radiol.2020191145

    work page doi:10.1148/radiol.2020191145 2020

  25. [25]

    A Guide to ComBat Harmonization of Imaging Biomarkers in Multicenter Studies

    Fanny Orlhac et al. “A Guide to ComBat Harmonization of Imaging Biomarkers in Multicenter Studies”. In:Journal of Nuclear Medicine63.2 (Feb. 2022), pp. 172–179.issn: 0161-5505.doi: 10.2967/jnumed.121.262464

    work page doi:10.2967/jnumed.121.262464 2022

  26. [26]

    Regression Shrinkage and Selection via the Lasso

    Robert Tibshirani. “Regression Shrinkage and Selection via the Lasso”. In:Journal of the Royal Statistical Society. Series B (Methodological)58.1 (1996), pp. 267–288.issn: 00359246

    work page 1996

  27. [27]

    Prostate cancer screening with prostate-specific antigen (PSA) test: a systematic review and meta-analysis

    Dragan Ilic et al. “Prostate cancer screening with prostate-specific antigen (PSA) test: a systematic review and meta-analysis”. In:BMJ362 (2018).issn: 0959-8138.doi:10.1136/ bmj.k3519

    work page 2018

  28. [28]

    Diagnostic accuracy of multi-parametric MRI and TRUS biopsy in prostate cancer (PROMIS): a paired validating confirmatory study

    Hashim U Ahmed et al. “Diagnostic accuracy of multi-parametric MRI and TRUS biopsy in prostate cancer (PROMIS): a paired validating confirmatory study”. In:The Lancet389.10071 (2017), pp. 815–822.issn: 0140-6736.doi:10.1016/S0140-6736(16)32401-1

    work page doi:10.1016/s0140-6736(16)32401-1 2017

  29. [29]

    F-18 labelled PSMA-1007: biodistribution, radiation dosimetry and histopathological validation of tumor lesions in prostate cancer patients

    Giesel F.L. et al. “F-18 labelled PSMA-1007: biodistribution, radiation dosimetry and histopathological validation of tumor lesions in prostate cancer patients”. In:European Journal of Nuclear Medicine and Molecular Imaging44 (2017), pp. 678–688.doi:10.1007/s00259- 016-3573-4

    work page doi:10.1007/s00259- 2017

  30. [30]

    A fully automatic method for biological target volume segmentation of brain metastases

    Alessandro Stefano et al. “A fully automatic method for biological target volume segmentation of brain metastases”. In:International Journal of Imaging Systems and Technology26.1 (2016), pp. 29–37.doi:10.1002/ima.22154

    work page doi:10.1002/ima.22154 2016

  31. [31]

    matRadiomics: A Novel and Complete Radiomics Framework, from Image Visualization to Predictive Model

    Giovanni Pasini et al. “matRadiomics: A Novel and Complete Radiomics Framework, from Image Visualization to Predictive Model”. In:Journal of Imaging8.8 (2022).issn: 2313-433X. doi:10.3390/jimaging8080221

    work page doi:10.3390/jimaging8080221 2022

  32. [32]

    Computational Radiomics System to Decode the Radio- graphic Phenotype

    Joost J.M. van Griethuysen et al. “Computational Radiomics System to Decode the Radio- graphic Phenotype”. In:Cancer Research77.21 (Oct. 2017), e104–e107.issn: 0008-5472.doi: 10.1158/0008-5472.CAN-17-0339. 15

    work page doi:10.1158/0008-5472.can-17-0339 2017