Pith. sign in

REVIEW 5 major objections 6 minor 33 references

Application of Quantum Convolutional Neural Networks for MRI-Based Brain Tumor Detection and Classification

T0 review · 5 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

Pith's one-line read A compact four-qubit quantum convolutional neural network classifies brain MRI scans into tumor and non-tumor with up to 89% accuracy, and into four tumor categories with up to 62% accuracy.

desk verdict The binary accuracy numbers are arithmetically incompatible with the paper's own test split, so the headline result collapses; the rest is a modest QCNN application write-up. read the letter →

arxiv 2509.02582 v1 pith:TGYV42FR submitted 2025-08-28 physics.med-ph cs.CV

classification physics.med-phcs.CV
keywords braintumorclassificationquantumconvolutionalneuralnetworkMRIimagebinarymulticlassoversamplingmachinelearningmedicalimaging
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper reports a proof-of-concept application of a compact quantum convolutional neural network (QCNN) to brain MRI classification. On a public dataset of 3,264 scans, a four-qubit QCNN with one quantum convolutional layer separates tumor from non-tumor images with 88–89% accuracy, and separates glioma, meningioma, pituitary, and non-tumor images with 52–62% accuracy. The central claim is that QCNNs are applicable to real medical imaging data, not just benchmark image sets, while using far fewer trainable parameters than a classical CNN. The binary result is the main evidence; the multiclass result is the acknowledged weak point that the authors attribute to circuit simplicity and limited data.

What carries the argument

The object that carries the argument is the QCNN model itself: a quantum convolutional layer with filter size 2 and depth 4, run on 4 qubits, followed by flatten and dense layers. The convolution operation is implemented as a sequence of controlled-NOT and controlled-Z gates applied iteratively over image patches, so the quantum circuit replaces the classical convolution kernel. The authors' claim is that this layer extracts sufficient features for binary tumor detection while keeping the trainable parameter count small; the same layer, with only CX and CZ entanglement, is also where they locate the multiclass model's limitation.

What would settle it

Compute the accuracy, sensitivity, and specificity of the trivial classifier that labels every test image as tumor on the same 653-image split. If the QCNN's accuracy is not materially above that ~84% baseline, or if its correct calls are concentrated in the tumor class while non-tumor recall stays near chance, the central claim of strong binary classification collapses.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a four-qubit QCNN—built from a quantum convolutional layer using controlled-NOT and controlled-Z gates, a flattening layer, and two dense layers—can be trained on 12x12 MRI thumbnails and produce working classifiers. The binary model reaches 88% accuracy (89% with oversampling), with the paper reporting that oversampling lifts sensitivity from 67% to 80% while trading precision down from 83% to 79%. The multiclass model reaches 52% accuracy, rising to 62% with oversampling. The authors interpret the binary performance as showing feasibility for automated MRI screening and the multiclass gap as evidence that the fixed filter size, shallo

Load-bearing premise

The load-bearing premise is that the 88–89% binary accuracy comes from learned tumor detection, not from the test set's class imbalance; since 550 of 653 test images are tumors, a classifier that always answers 'tumor' would already be about 84% accurate, and the paper never reports that baseline.

Editorial extensions

If this is right

  • A four-qubit QCNN can separate tumor from non-tumor MRI images at 88–89% accuracy on a public dataset, supporting the paper's claim that compact quantum models are usable for automated screening.
  • Because the QCNN uses far fewer trainable parameters than a classical CNN, the result implies a low-parameter route to MRI classification that is practical to train in simulation.
  • Oversampling the minority class changes the trade-off: binary sensitivity rises from 67% to 80%, precision falls from 83% to 79%, indicating that the headline accuracy gain comes with a different error profile.
  • Multiclass tumor typing at 62% accuracy is below what the paper itself considers clinically usable, so the result is a call for deeper quantum circuits and richer entanglement before multiclass QCNN diagnosis is viable.
  • The authors' stated next steps—optimized circuit architectures, hybrid classical-quantum training, and larger datasets—follow directly from the observed multiclass gap.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The 88–89% binary accuracy should be read against the trivial rule 'always predict tumor', which would score roughly 84% on the same test split because 550 of 653 test images are tumor; the paper does not report this baseline, so the true learned gain over a no-information classifier is at most a few points.
  • The multiclass oversampling result is trained on duplicated minority samples, so validation accuracy may overstate performance on the original class distribution; reporting per-class recall on the unmodified test set would settle this.
  • The paper's introduction points to a direct comparison with a lightweight classical CNN under the same split, but the results section reports only QCNN metrics; until that comparison appears, the 'comparable performance with fewer parameters' claim remains an assertion, not a demonstrated trade-off.
  • If a deeper or more entangled quantum circuit fails to improve the 62% multiclass ceiling on the same data, the bottleneck is likely the quantum feature extractor's expressiveness rather than dataset size, which is a testable prediction for future work.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper proposes a compact four-qubit Quantum Convolutional Neural Network (QCNN) for brain tumor detection and classification from MRI, using a single quantum convolutional layer with CX/CZ gates, depth 4, filter size 2, and a downstream dense head. Two tasks are reported: binary tumor/non-tumor classification and four-class tumor-type classification, both on an 80/20 split of a 3,264-image Kaggle dataset. The authors report binary accuracy 88% (89% after oversampling) and multiclass accuracy 52% (62% after oversampling), and interpret the binary result as strong and promising for medical imaging, while acknowledging multiclass limitations. The paper also claims, in the introduction and discussion, that the QCNN provides a direct comparison to a lightweight classical CNN and achieves comparable performance with fewer parameters and shorter training time, although no such comparison is reported.

Significance. If the reported results were valid, the paper would provide one of the first demonstrations of a QCNN applied directly to clinical-style brain MRI rather than generic benchmarks such as MNIST. The compact four-qubit architecture and the explicit focus on a real medical imaging task are potentially useful contributions. However, the paper's central quantitative claim is undermined by an internal arithmetic inconsistency in the binary validation metrics, and the manuscript provides neither a classical baseline nor a majority-class baseline, making the significance of the headline numbers impossible to assess as written. The multiclass results are above chance (25%) but modest, and the absence of any uncertainty or repeated-run statistics further limits the strength of the conclusions.

major comments (5)
  1. [Tables 2 and 4] The binary validation metrics reported in Table 4 are not simultaneously realizable on the validation split described in Table 2. Table 2 states 550 tumor-positive and 103 tumor-negative validation images (653 total). For the original binary model, accuracy 88% implies about 575 correct predictions, while sensitivity 67% implies TP ≈ 0.67×550 ≈ 368, requiring TN ≈ 575 − 368 ≈ 207, which exceeds the 103 available negatives. For the oversampled row, accuracy 89% and sensitivity 80% imply TP = 440 and TN ≈ 581 − 440 = 141 > 103. No integer confusion matrix with TP+FN=550 and TN+FP=103 can reproduce either pair. The paper must report the actual confusion matrices and clarify whether the metrics were computed on a differently composed validation set. As written, the headline binary accuracy numbers are not supported.
  2. [Introduction and Discussion (unlabeled comparison section)] The manuscript repeatedly states that the QCNN is compared with a lightweight classical CNN under the same data split and that it achieves comparable performance with far fewer parameters and shorter training time. No such experiment, table, or result appears anywhere in Section 3 or elsewhere. Without a classical CNN baseline, the claim of 'comparable performance' is unsupported, and the paper cannot substantiate any quantum-vs-classical advantage or trade-off. Either the comparison must be reported, or these statements must be removed.
  3. [Table 2 and Section 3 'Binary classification model'] No majority-class baseline is reported. With 550 positive and 103 negative validation images, a trivial classifier that always predicts 'tumor' attains 84.2% accuracy. The reported 88–89% accuracy is only marginally above this baseline, and the sensitivity of 67% before oversampling is below the trivial classifier's 100% sensitivity. The paper calls the binary performance 'strong' without acknowledging this baseline. Even after the arithmetic issue is corrected, the authors should report sensitivity, specificity, and majority-class accuracy to qualify the claim.
  4. [Sections 2.2, 2.3, and Table 4] The use of oversampling is described inconsistently. The abstract says oversampling was applied to address class imbalance; Section 2.3 states it was performed only on the binary dataset; but Table 4 reports 'Multiclass classification with oversampling', and Section 3 describes it as a separate experiment applied to the multiclass task. The Methods provide no procedure for oversampling the multiclass set, and it is unclear how the test/validation split was treated under duplication. This must be clarified before the multiclass oversampling result can be reproduced.
  5. [Section 3 and Table 4] All reported metrics are single point estimates from a single training run. No random seed, cross-validation, repeated runs, standard deviation, or confidence interval is provided. Given the modest differences (52% vs 62%; 88% vs 89%) and the absence of any uncertainty quantification, the claim that oversampling 'improved' the models cannot be statistically evaluated. The authors should provide repeated-run statistics or at least state the stochastic variations across runs.
minor comments (6)
  1. [Equations (1) and (2)] The matrix representations of CX and CZ are not formatted as standard 4×4 matrices. The current notation is ambiguous and likely contains typesetting errors; please replace with standard matrix forms.
  2. [Section 2.2] The text introduces 'four filters used are convolution filters consisting of 3 classical convolution filters and 1 quantum convolution filter' without defining the classical filters or how they interact with the quantum layer. This is needed for reproducibility.
  3. [Figures 3 and 4] The text uses 'Figure 3' for both the multiclass and binary model visualizations. The binary model should refer to Figure 4. Please correct the cross-references.
  4. [Section 3 'Multiclass classification'] The manuscript says 'Comparing both graphs portrays opposite trends', but only one epoch-versus-accuracy graph is shown. Clarify what the second graph is (presumably loss) and ensure both are included.
  5. [Section 3 'Multiclass classification'] The statement 'improving the accuracy of the model by 20% in the data training process and 10% in the validation process' conflates percentage points and relative percentages. The table shows 52%→62%, a 10 percentage-point gain. Please use unambiguous phrasing.
  6. [Section 2.1] The parenthetical '(noting a correction from the original 3,160 to align with earlier manuscript details)' is unclear and appears to refer to manuscript revision history rather than a scientific detail. Remove or rephrase.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: reported accuracies are direct held-out validation measurements, not quantities derived from model inputs or self-citations.

full rationale

This paper contains no derivation chain in which an output is built into an input. The QCNN configuration (filter size 2, depth 4, 4 qubits, CX/CZ gates) is stated as an architectural choice, not derived from the reported accuracies. The binary 88%/89% and multiclass 52%/62% figures are presented as confusion-matrix metrics computed on a fixed 20% validation split (Tables 2 and 4). No metric is used as a fitting target, no self-citation carries a load-bearing assumption (prior QCNN references are background, not justification for the reported numbers), and no uniqueness theorem or ansatz is imported from the authors' own prior work. The paper's limitation statements (small dataset, shallow circuit, single quantum convolutional layer) are explicit and do not conceal a circular step. Two non-circular validity concerns should be kept separate: (1) the paper never reports the majority-class baseline of 84.2% (550/653) against which the binary 88-89% accuracy should be judged; (2) the binary Table 4 metrics appear arithmetically incompatible with the 550/103 validation split stated in Table 2. Both are correctness/reporting issues, not instances of fitted parameters being renamed as predictions.

Assumptions & free parameters 6 free parameters · 6 assumptions · 0 invented entities

The paper is an empirical application, not a derivation, so the ledger consists of hand-chosen hyperparameters and unstated implementation assumptions. No new physical entities are introduced. The lack of a specified encoding circuit, simulator, optimizer, and code means the central numbers rest on choices the reader cannot reproduce from the text.

free parameters (6)
  • convolution_filter_size = 2
    Hand-chosen; sets qubit count via qubit = filter_size^2 and determines the receptive field of the quantum convolution.
  • circuit_depth = 4
    Hand-chosen; number of sequential CX/CZ convolution passes over the image.
  • num_qubits = 4
    Consequence of filter size 2; not independently tuned.
  • training_epochs = 10
    All models stopped at 10 epochs; no validation-based early stopping or longer training reported.
  • image_resize = 12x12 pixels
    All MRI images downsampled to 12x12 before encoding, discarding most spatial detail.
  • oversampling_ratio = minority class duplicated from 397 to 2214 samples
    Random oversampling copies non-tumor training images to match the tumor majority; a separate multiclass oversampling experiment is reported despite Section 2.2 saying oversampling was only applied to the binary dataset.
assumptions (6)
  • standard math CX and CZ gate operations are unitary and reversible as defined in Eqs. (1) and (2).
    The quantum convolution layer is built entirely from these gates.
  • domain assumption The Kaggle MRI dataset labels are correct and the 80:20 split gives a meaningful generalization estimate.
    All accuracy claims are measured against this assumed ground truth; no label audit is performed.
  • domain assumption Images can be encoded into 4 qubits by a scheme tied to the filter size and repeated over the image.
    The exact encoding circuit is not specified, but the model's expressiveness depends entirely on it.
  • domain assumption A classical simulation of the 4-qubit circuit faithfully represents the intended quantum computation.
    The paper never names the simulator or backend and reports no noise model, yet generalizes to near-term quantum devices in the discussion.
  • domain assumption Random oversampling by duplicating minority samples does not invalidate the evaluation.
    The validation set remains imbalanced after balancing the training set, and the effect of duplicated training data is not discussed.
  • ad hoc to paper A single quantum convolutional layer with fixed CX/CZ entanglement is expressive enough to support the classification claims.
    This architecture is chosen without ablations or circuit search; the paper attributes weak multiclass performance to this same design.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Application of Quantum Convolutional Neural Networks for MRI-Based Brain Tumor Detection and Classification." pith.science (2026). https://pith.science/paper/TGYV42FR

@misc{pith2026250902582,
  author       = {Pith},
  title        = {Pith review of: Application of Quantum Convolutional Neural Networks for MRI-Based Brain Tumor Detection and Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TGYV42FR}},
  note         = {Machine review of arXiv:2509.02582}
}
read the original abstract

This study explores the application of Quantum Convolutional Neural Networks (QCNNs) for brain tumor classification using MRI images, leveraging quantum computing for enhanced computational efficiency. A dataset of 3,264 MRI images, including glioma, meningioma, pituitary tumors, and non-tumor cases, was utilized. The data was split into 80% training and 20% testing, with an oversampling technique applied to address class imbalance. The QCNN model consists of quantum convolution layers, flatten layers, and dense layers, with a filter size of 2, depth of 4, and 4 qubits, trained over 10 epochs. Two models were developed: a binary classification model distinguishing tumor presence and a multiclass classification model categorizing tumor types. The binary model achieved 88% accuracy, improving to 89% after data balancing, while the multiclass model achieved 52% accuracy, increasing to 62% after oversampling. Despite strong binary classification performance, the multiclass model faced challenges due to dataset complexity and quantum circuit limitations. These findings suggest that QCNNs hold promise for medical imaging applications, particularly in binary classification. However, further refinements, including optimized quantum circuit architectures and hybrid classical-quantum approaches, are necessary to enhance multiclass classification accuracy and improve QCNN applicability in clinical settings.

Figures

Figures reproduced from arXiv: 2509.02582 by the authors.

Figure 1
Figure 1. MRI Image Data of Patients with Glioma Brain Tumors (First Row), Meningioma [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Confusion matrix illustrating TP, FP, TN, and FN values The confusion matrix can be used for accuracy, sensitivity, precision, and F1-score ( [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 22 canonical work pages

  1. [1]

    Clinical presentation and epidemiology of brain tumors firstly diagnosed in adults in the Emergency Department: a 10- year, single center retrospective study

    Comelli I, Lippi G, Campana V, Servadei F, Cervellin G. Clinical presentation and epidemiology of brain tumors firstly diagnosed in adults in the Emergency Department: a 10- year, single center retrospective study. Ann Transl Med 2017;5:269. https://doi.org/10.21037/atm.2017.06.12

  2. [2]

    Clinical features of primary brain tumours: a case –control study using electronic primary care records

    Hamilton W, Kernick D. Clinical features of primary brain tumours: a case –control study using electronic primary care records. Br J Gen Pract 2007;57:695–9

  3. [3]

    Psychiatric aspects of brain tumors: A review

    Madhusoodanan S, Ting MB, Farah T, Ugur U. Psychiatric aspects of brain tumors: A review. World J Psychiatry 2015;5:273–85. https://doi.org/10.5498/wjp.v5.i3.273

  4. [4]

    Brain Tumors: Epidemiology and Current Trends in Treatment

    Strong M, Garces J. Brain Tumors: Epidemiology and Current Trends in Treatment. J Brain Tumors Neurooncology 2016;01. https://doi.org/10.4172/2475-3203.1000102

  5. [5]

    Detection and isolation of brain tumors in cancer patients using neural network techniques in MRI images

    Mir M, Madhi ZS, Hamid AbdulHussein A, Khodayer Hassan Al Dulaimi M, Suliman M, Alkhayyat A, et al. Detection and isolation of brain tumors in cancer patients using neural network techniques in MRI images. Sci Rep 2024;14:23341. https://doi.org/10.1038/s41598- 024-68567-5

  6. [6]

    Current Clinical Brain Tumor Imaging

    Villanueva-Meyer JE, Mabray MC, Cha S. Current Clinical Brain Tumor Imaging. Neurosurgery 2017;81:397–415. https://doi.org/10.1093/neuros/nyx103

  7. [7]

    MRI -powered biomedical devices

    Hovet S, Ren H, Xu S, Wood B, Tokuda J, Tse ZTH. MRI -powered biomedical devices. Minim Invasive Ther Allied Technol MITAT Off J Soc Minim Invasive Ther 2018;27:191 –

  8. [8]

    Magnetic Resonance Imaging of Primary Adult Brain Tumors: State of the Art and Future Perspectives

    Martucci M, Russo R, Schimperna F, D’Apolito G, Panfili M, Grimaldi A, et al. Magnetic Resonance Imaging of Primary Adult Brain Tumors: State of the Art and Future Perspectives. Biomedicines 2023;11:364. https://doi.org/10.3390/biomedicines11020364

Show all 33 references
  1. [9]

    Machine learning and transfer learning techniques for accurate brain tumor classification

    Matin Malakouti S, Bagher Menhaj M, Abolfazl Suratgar A. Machine learning and transfer learning techniques for accurate brain tumor classification. Clin EHealth 2024;7:106 –19. https://doi.org/10.1016/j.ceh.2024.08.001

  2. [10]

    MRI Image Based Alzheimer’s Disease Classification Using Convolutional Neural Network: EfficientNet Architecture

    Ujilast NA, Firdausita NS, Aditya CSK, Azhar Y. MRI Image Based Alzheimer’s Disease Classification Using Convolutional Neural Network: EfficientNet Architecture. J RESTI Rekayasa Sist Dan Teknol Inf 2024;8:18–25. https://doi.org/10.29207/resti.v8i1.5457

  3. [11]

    MRI Brain Tumor Image Classification Using a Combined Feature and Image -Based Classifier

    Veeramuthu A, Meenakshi S, Mathivanan G, Kotecha K, Saini JR, Vijayakumar V, et al. MRI Brain Tumor Image Classification Using a Combined Feature and Image -Based Classifier. Front Psychol 2022;13. https://doi.org/10.3389/fpsyg.2022.848784

  4. [12]

    Modernizing Neuro- Oncology: The Impact of Imaging, Liquid Biopsies, and AI on Diagnosis and Treatment

    Rafanan J, Ghani N, Kazemeini S, Nadeem-Tariq A, Shih R, Vida TA. Modernizing Neuro- Oncology: The Impact of Imaging, Liquid Biopsies, and AI on Diagnosis and Treatment. Int J Mol Sci 2025;26:917. https://doi.org/10.3390/ijms26030917

  5. [13]

    Automated brain tumor identification using magnetic resonance imaging: A systematic review and meta - analysis

    Kouli O, Hassane A, Badran D, Kouli T, Hossain -Ibrahim K, Steele JD. Automated brain tumor identification using magnetic resonance imaging: A systematic review and meta - analysis. Neuro-Oncol Adv 2022;4:vdac081. https://doi.org/10.1093/noajnl/vdac081

  6. [14]

    Deep learning models for CT image classification: a comprehensive literature review

    Ahmad IS, Dai J, Xie Y, Liang X. Deep learning models for CT image classification: a comprehensive literature review. Quant Imaging Med Surg 2025;15:962011 –961011. https://doi.org/10.21037/qims-24-1400

  7. [15]

    Review of Image Classification Algorithms Based on Convolutional Neural Networks

    Chen L, Li S, Bai Q, Yang J, Jiang S, Miao Y. Review of Image Classification Algorithms Based on Convolutional Neural Networks. Remote Sens 2021;13:4712. https://doi.org/10.3390/rs13224712

  8. [16]

    MRI -based brain tumour image detection using CNN based deep learning method

    Chattopadhyay A, Maitra M. MRI -based brain tumour image detection using CNN based deep learning method. Neurosci Inform 2022;2:100060. https://doi.org/10.1016/j.neuri.2022.100060

  9. [17]

    A novel CNN architecture for accurate early detection and classification of Alzheimer’s disease using MRI data

    El-Assy AM, Amer HM, Ibrahim HM, Mohamed MA. A novel CNN architecture for accurate early detection and classification of Alzheimer’s disease using MRI data. Sci Rep 2024;14:3463. https://doi.org/10.1038/s41598-024-53733-6

  10. [18]

    Niakan Kalhori S

    Saeedi S, Rezayi S, Keshavarz H, R. Niakan Kalhori S. MRI -based brain tumor detection using convolutional deep learning methods and chosen machine learning techniques. BMC Med Inform Decis Mak 2023;23:16. https://doi.org/10.1186/s12911-023-02114-6

  11. [19]

    Multiclass classification using quantum convolutional neural networks with hybrid quantum-classical learning

    Bokhan D, Mastiukova AS, Boev AS, Trubnikov DN, Fedorov AK. Multiclass classification using quantum convolutional neural networks with hybrid quantum-classical learning. Front Phys 2022;10:1069985. https://doi.org/10.3389/fphy.2022.1069985

  12. [20]

    Quantum convolutional neural network for image classification

    Chen G, Chen Q, Long S, Zhu W, Yuan Z, Wu Y. Quantum convolutional neural network for image classification. Pattern Anal Appl 2023;26:655–67. https://doi.org/10.1007/s10044- 022-01113-z

  13. [21]

    Quantum convolutional neural networks for jet images classification 2025

    Elhag H, Hartung T, Jansen K, Nagano L, Pirina GM, Tucci AD. Quantum convolutional neural networks for jet images classification 2025. https://doi.org/10.48550/arXiv.2408.08701

  14. [22]

    Quantum convolutional neural network for classical data classification

    Hur T, Kim L, Park DK. Quantum convolutional neural network for classical data classification. Quantum Mach Intell 2022;4:3. https://doi.org/10.1007/s42484-021-00061-x

  15. [23]

    Scalable quantum convolutional neural network for image classification

    Sun Y, Li D, Xiang Q, Yuan Y, Hu Z, Hua X, et al. Scalable quantum convolutional neural network for image classification. Phys Stat Mech Its Appl 2025;657:130226. https://doi.org/10.1016/j.physa.2024.130226

  16. [24]

    A Tutorial on Quantum Convolutional Neural Networks (QCNN) 2020

    Oh S, Choi J, Kim J. A Tutorial on Quantum Convolutional Neural Networks (QCNN) 2020. https://doi.org/10.48550/arXiv.2009.09423

  17. [25]

    Quanvolutional neural networks: powering image recognition with quantum circuits

    Henderson M, Shakya S, Pradhan S, Cook T. Quanvolutional neural networks: powering image recognition with quantum circuits. Quantum Mach Intell 2020;2:2. https://doi.org/10.1007/s42484-020-00012-y

  18. [26]

    Quantum convolutional neural networks for multiclass image classification

    Shi S, Wang Z, Li J, Li Y, Shang R, Zhong G, et al. Quantum convolutional neural networks for multiclass image classification. Quantum Inf Process 2024;23:189. https://doi.org/10.1007/s11128-024-04360-7

  19. [27]

    Breast Cancer Detection with Quanvolutional Neural Networks

    Matondo-Mvula N, Elleithy K. Breast Cancer Detection with Quanvolutional Neural Networks. Entropy 2024;26:630. https://doi.org/10.3390/e26080630

  20. [28]

    Quantum classical hybrid convolutional neural networks for breast cancer diagnosis

    Xiang Q, Li D, Hu Z, Yuan Y, Sun Y, Zhu Y, et al. Quantum classical hybrid convolutional neural networks for breast cancer diagnosis. Sci Rep 2024;14:24699. https://doi.org/10.1038/s41598-024-74778-7

  21. [29]

    A Distributed Hybrid Quantum Convolutional Neural Network for Medical Image Classification 2025

    Li Y, Qia Z, Lia Y, Yanga H, Shanga R, Jiaoa L. A Distributed Hybrid Quantum Convolutional Neural Network for Medical Image Classification 2025. https://doi.org/10.48550/arXiv.2501.06225

  22. [30]

    https://www.kaggle.com/datasets/sartajbhuvaji/brain- tumor-classification-mri (accessed April 1, 2025)

    Brain Tumor Classification (MRI) n.d. https://www.kaggle.com/datasets/sartajbhuvaji/brain- tumor-classification-mri (accessed April 1, 2025)

  23. [31]

    Investigating the Impact of Train / Test Split Ratio on the Performance of Pre -Trained Models with Custom Datasets

    Bichri H, Chergui A, Hain M. Investigating the Impact of Train / Test Split Ratio on the Performance of Pre -Trained Models with Custom Datasets. Int J Adv Comput Sci Appl 2024;15. https://doi.org/10.14569/IJACSA.2024.0150235

  24. [32]

    Machine Learning with Oversampling and Undersampling Techniques: Overview Study and Experimental Results

    Mohammed R, Rawashdeh J, Abdullah M. Machine Learning with Oversampling and Undersampling Techniques: Overview Study and Experimental Results. 2020 11th Int. Conf. Inf. Commun. Syst. ICICS, 2020, p. 243 –8. https://doi.org/10.1109/ICICS49469.2020.239556

  25. [202]

    https://doi.org/10.1080/13645706.2017.1402188

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.