The DOI matches its Crossref record; 36 of the 40 references it cites were found in a registry or an index; 10 of the 12 sources its claims cite were checked for support.
Medical image diagnosis based on adaptive Hybrid Quantum CNN
Open accessPublished version of a preprint
Abstract
Hybrid quantum systems have shown promise in image classification by combining the strengths of both classical and quantum algorithms. These systems leverage the parallel processing power of quantum computers to perform complex computations while utilizing classical algorithms to handle the vast amounts of data involved in imaging. The hybrid approach is intended to improve accuracy and speed compared to traditional classical methods. Further research and development in this area can revolutionize the way medical images are classified and help improve patient diagnosis and treatment. The use of Conventional Neural Networks (CNN) for the classification and diagnosis of medical images using big datasets requires, in most cases, the use of special high-performance computing machines, which are very expensive and hard to access by most researchers. A new form of Machine Learning (ML), Quantum machine learning (QML), is being introduced as an emerging strategy to overcome this problem. A hybrid quantum-classical CNN uses both quantum and classical convolution layers designed to use a parameterized quantum circuit. This means that the computing model utilizes a quantum circuits approach to construct QML algorithms, which are then used to transform the quantum state to extract image hidden features. This computational acceleration is expected to achieve better algorithm performance than classical CNNs. This study intends to evaluate the performance of a Hybrid Quantum CNN (HQCNN) against a conventional CNN. This is followed by some optimizer modifications for both proposed and classical CNN methods to investigate the possible further improvement of their performance. The optimizer modification is based on forcing the optimizer to be directly adaptive to model accuracy. The optimizer adaptiveness is based on the development of an optimizer with a loss base adaptive momentum. Several algorithms are developed to achieve the above-mentioned goals, including CNN, QCNN, CNN with the adaptive optimizer, and QCNN with the Adaptive optimizer. The four algorithms are tested against a Kaggle brain dataset containing over 7000 samples. The test results show the hybrid quantum circuit algorithm outperformed the conventional CNN algorithm. The performance of both algorithms was further improved by using a fully adaptive SGD optimizer.
Overview
CoChat Summary
## TL;DR
The paper proposes a hybrid quantum-classical convolutional neural network (HQCNN) for brain tumor image classification and compares it with a conventional CNN on two MRI datasets. It reports that HQCNN converges faster and, with a fully adaptive SGD optimizer, reaches 98.07% accuracy within 70 epochs, outperforming the CNN.
## Key findings
- On the Kaggle Brain Tumor MRI dataset, the study reports 98.07% accuracy for the HQCNN within 70 epochs, while the conventional CNN needed 100 epochs for nearly the same accuracy level.
- The results section says the quantum circuit model reached the same accuracy as the CNN within 70 epochs, and the HQCNN reached it within 63 epochs.
- The paper reports that the fully adaptive SGD optimizer improved both the classical CNN and the quantum circuit model, improving convergence speed.
- Statistical testing in the results reports p = 0.0027 for a paired t-test, which the paper interprets as a statistically significant performance difference; it also reports Cohen’s d = 0.184 at epoch 1 and Cohen’s d = 2.385 at epoch 20.
## Methods
- The study compares CNN, HQCNN, and versions with a fully adaptive SGD optimizer; it frames the HQCNN as combining a quantum convolutional layer, quantum pooling, and a classical fully connected layer.
- Two datasets were used: the Kaggle Brain Tumor MRI dataset with 7,023 images across 4 classes and the REM-BRANDT dataset with 106,541 DICOM images after preprocessing, also split into 4 classes.
- Model structure was optimized using Genetic Algorithms (GA); performance was evaluated with accuracy, recall, precision, and F1 score, along with t-tests and confidence intervals.
## Limitations
- The paper explicitly says further investigation is warranted to assess performance on other types of data.
- It also states future work is needed to study robustness against adversarial attacks.
- Evident from the text: several sections in the provided record are incomplete or truncated, and some equation references and figures/tables are mentioned but not fully shown; this limits what can be verified from the text alone.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures














Verification
Claims the paper makes
1 Valid · 2 Not valid · 1 Not checked · 44 Own results and interpretations
Some cited sources support this claim and some don’t — read each source’s verdict below.
In the paper
“Several QML algorithms based on PQC have been proposed [21–27], demonstrating superior performance and computational abilities compared to conventional algorithms [28–31].”
§ Medical image diagnosis based on adaptive Hybrid Quantum CNN > Introduction
What the sources say
- Partly supportedcites [21] Wecker D, Hastings MB, Troyer M. Progress towards practical quantum variational algorithms. Phys Rev A. 2015;92(4):042303. sourceExperimental — accuracy not yet measured
The opening summary (p. 1) reports that a parameterized quantum-circuit ansatz converged faster than earlier unitary-coupled-cluster schemes. The Discussion (p. 10) describes outperforming classical computers as a prospect, not a demonstrated result. The paper therefore does not establish the cited claim that several quantum machine-learning algorithms have demonstrated superior performance over conventional algorithms. (LLM, conf=0.86)
- Partly supportedcites [22] Jones T, Endo S, McArdle S, Yuan X, Benjamin SC. Variational quantum algorithms for discovering Hamiltonian spectra. Phys Rev A. 2019;99(6):062304. sourceExperimental — accuracy not yet measured
The Discussion reports that the proposed variational quantum algorithm estimated spectra in simulations and that, in those simulations, gradient descent was more prone to local minima. The paper does not demonstrate superior performance or computational ability over conventional algorithms generally, nor does it present its method as a quantum machine-learning algorithm. (LLM, conf=0.84)
- Not checked — no full text of the sourcecites [23] Xiao J, Wen J, Wei S, Long G. Reconstructing unknown quantum states using variational layerwise method. Front Phys. 2022;17(5):51501. source
- Disputedcites [24] Guerreschi GG, Matsuura AY. QAOA for Max-Cut requires hundreds of qbits for quantum speed-up. Sci Rep. 2019;9(1):6903. sourceExperimental — accuracy not yet measured
The Results section (“Comparison and performance crossover,” page unavailable) says the classical solver is much faster for the tested small instances. The Discussion estimates that QAOA might achieve a speedup only at a crossover involving several hundred to a few thousand qubits. The paper therefore does not demonstrate superior performance by PQC-based QML algorithms, as the cited clause claims. (LLM, conf=0.86)
Checked by a reasoning model — may be wrong.
- Partly supportedcites [25] Zhou L, Wang ST, Choi S, Pichler H, Lukin MD. Quantum approximate optimization algorithm: performance, mechanism, and implementation on near-term devices. Phys Rev X. 2020;10(2):021067. sourceExperimental — accuracy not yet measured
The Introduction describes QAOA as a variational-circuit algorithm with performance guarantees and states that it is not efficiently simulatable by classical computers. But Section II calls an advantage over classical algorithms *potential*, and the large-problem implementation section says it remains to be seen whether QAOA can outperform a classical MaxCut algorithm. The paper therefore does not establish the claimed demonstrated superiority over conventional algorithms. (LLM, conf=0.87)
- Partly supportedcites [26] Rasmussen SE, Loft NJS, Bækkegaard T, Kues M, Zinner NT. Reducing the amount of single-qbit rotations in VQE and related algorithms. Advanced Quantum Technol. 2020;3(12):2000063. sourceExperimental — accuracy not yet measured
The Introduction (p. 1) of Rasmussen et al. does enumerate several hybrid quantum-classical / machine-learning algorithms built on parameterized quantum circuits (QAOA, QAE, QVECTOR, classification via near-term QNN, qGAN, VQE), which supports the 'several QML algorithms based on PQC have been proposed' half of the claim. However, no passage in our copy states or implies that these algorithms have demonstrated superior performance or computational abilities compared to conventional/classical algorithms; instead the paper frames NISQ devices as noise-limited (Introduction, p. 1: 'a couple of tens of qubits and a few hundred gate operations before the noise becomes too overwhelming') and treats the classical optimization load as a probable limiting factor, aiming to reduce it (Conclusion, p. 10). The superiority-over-conventional-algorithms part attributed to this source is therefore not supported. (LLM, conf=0.75)
- Partly supportedcites [27] Uvarov AV, Kardashin AS, Biamonte JD. Machine learning phase transitions with a quantum processor. Phys Rev A. 2020;102(1):012415. sourceExperimental — accuracy not yet measured
Our copy supports an asserted advantage over classical methods but not a demonstrated head-to-head comparison. The Conclusions (p. 5) state that 'the quantum classifier works intrinsically with quantum data, providing advantage over the classical methods based on Monte-Carlo sampling' because classical manipulation of the states 'would require an exponential amount of memory in the worst case', and the Introduction (p. 1) says only that quantum simulators 'are predicted to simulate quantum systems efficiently' and that 'there is no known classical algorithm to efficiently implement this method' (p. 2). The paper's own results are noiseless Qiskit simulations of a single proposed variational classifier (99% TFIM, 94% XXZ accuracy, pp. 1, 4), and the authors explicitly note that 'it would be interesting to compare our model with a classical model' based on MPS/MERA states (p. 5) — i.e., no classical ML baseline was run. So the part of the claim that superior performance was 'demonstrated ... compared to conventional algorithms' is not supported; only a predicted/structural computational advantage is. (LLM, conf=0.72)
- Partly supportedcites [28] Liu J, Lim KH, Wood KL, Huang W, Guo C, Huang HL. Hybrid quantum-classical convolutional neural networks. Sci China Phys Mechanics Astronomy. 2021;64(9):290311. sourceExperimental — accuracy not yet measured
The Introduction (p. 1) does attribute to PQC-based QML algorithms 'two main potential advantages, stronger expressive power [24-29] and stronger computing power [30]' relative to classical neural networks, which covers the comparative framing of the claim. But the source hedges this as *potential* advantages originating from superposition, whereas the citing sentence asserts that several such algorithms 'have demonstrated superior performance and computational abilities'. The source's own discussion (IV, p. 6) even states that for its QCCNN 'from the point of computational efficiency QCCNN would not be advantageous', with quantum advantage in computational time only conditional on finding a problem free of the input problem; the only demonstrated superiority is higher classification accuracy on the synthetic Tetris dataset for its own model, not for several existing algorithms. (LLM, conf=0.78)
- Partly supportedcites [29] Schuld M, Killoran N. Quantum machine learning in feature hilbert spaces. Phys Rev Lett. 2019;122(9):290311. sourceExperimental — accuracy not yet measured
The Conclusion describes a variational quantum circuit classifier and promising small-scale simulations. But Section D says the demonstrated squeezing-based model can be efficiently simulated classically, while identifying classically intractable feature maps remains an open question. The paper therefore does not demonstrate superior performance or computational ability over conventional algorithms. (LLM, conf=0.86)
- Partly supportedcites [30] Wei S, Chen Y, Zhou Z, Long G. A quantum convolutional neural network on NISQ devices. AAPPS Bull. 2022;32(1):2. sourceExperimental — accuracy not yet measured
Wei et al. (2022) support the computational-ability half of the attribution: Section 3 (p. 7) shows the convolutional layer runs in O((log2 M)^6) basic gates versus O(M^2) for a classical CNN, and the Discussion (p. 9) summarises 'exponential speed-ups over their classical counterparts in gate complexity' with fewer variational parameters. They do not support 'superior performance': the Discussion states 'With fewer parameters, our model achieves similar performance compared with classical algorithm in handwritten number recognition tasks', and Table 1/Fig. 4 (Section 2.2.2, pp. 6-7) give the CNN slightly higher average accuracy (0.972 vs 0.963 test set for the 2-class task). Additionally, this paper is a single proposed QCNN, not a survey of 'several' PQC-based algorithms, so the plural generalisation exceeds what this source reports. (LLM, conf=0.80)
boostedstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The table also demonstrates that incorporating a fully adaptive SGD optimizer improved the performance of both the CNN and Quantum Circuit models.”
§ Medical image diagnosis based on adaptive Hybrid Quantum CNN > Simulation and results > Results and discussion
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“At Epoch 20, the p-value is 0.000, which is significantly lower than the significance threshold. This indicates a statistically significant difference in validation accuracy between the CNN and HQCNN models at this epoch.”
§ Medical image diagnosis based on adaptive Hybrid Quantum CNN > Simulation and results > Results and discussion
hedgedstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The HQCNN model exhibits superior performance compared to a conventional CNN model on the same dataset, achieving an impressive accuracy of 98.07% within just 70 epochs, surpassing the conventional CNN's performance that required 100 epochs for nearly the same accuracy level.”
§ Medical image diagnosis based on adaptive Hybrid Quantum CNN > Conclusion
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, it is important to note that when the CNN and Quantum circuit models are trained using a GPU, the processing time difference between the two models becomes comparable.”
§ Medical image diagnosis based on adaptive Hybrid Quantum CNN > Simulation and results > Results and discussion
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
23 more claims
Create a free account to see all 48 claims and every source verdict.
- Probably not a substantial claim. 9
- It may state two separate claims as one. 10
- The quote only partly states this claim. 2
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Medical image diagnosis based on adaptive Hybrid Quantum CNN > Simulation and results > Experimental design
“Therefore, the total number of trainable variables for the entire model is 8256.”
§ Medical image diagnosis based on adaptive Hybrid Quantum CNN > Simulation and results > Experimental design
“So, the total number of trainable variables in the hybrid quantum convolutional neural network is 9160.”
Citations
Cites
A variational eigenvalue solver on a photonic quantum processor
Peruzzo, A., McClean, J., Shadbolt, P., Yung, M.-H., Zhou, X.-Q., Love, P. J., Aspuru-Guzik, A., & O’Brien, J. L. (2014) · Nature Communications · doi:10.1038/ncomms5213 · Cited by 4,552 · cited 1×
“Notably, Peruzzo et al. introduced the variational quantum eigensolver (VQE) [17], and subsequent work by McClean et al. improved its optimization scheme [18].”
MethodQuantum computational advantage using photons
Zhong, H.-S., Wang, H., Deng, Y.-H., Chen, M.-C., Peng, L.-C., Luo, Y.-H., Qin, J., Wu, D., Ding, X., Hu, Y., et al. (2020) · Science · doi:10.1126/science.abe8770 · Cited by 2,405
The theory of variational hybrid quantum-classical algorithms
McClean, J. R., Romero, J., Babbush, R., & Aspuru-Guzik, A. (2016) · New Journal of Physics · doi:10.1088/1367-2630/18/2/023023 · Cited by 2,227
Quantum Support Vector Machine for Big Data Classification
Rebentrost, P., Mohseni, M., & Lloyd, S. (2014) · Physical Review Letters · doi:10.1103/physrevlett.113.130503 · Cited by 1,951
Quantum Machine Learning in Feature Hilbert Spaces
Schuld, M., & Killoran, N. (2019) · Physical Review Letters · doi:10.1103/physrevlett.122.040504 · Cited by 1,840
Quantum convolutional neural networks
Cong, I., Choi, S., & Lukin, M. D. (2019) · Nature Physics · doi:10.1038/s41567-019-0648-8 · Cited by 1,517
Circuit-centric quantum classifiers
Schuld, M., Bocharov, A., Svore, K. M., & Wiebe, N. (2020) · Physical Review A · doi:10.1103/physreva.101.032308 · Cited by 1,057
Zhou, L., Wang, S.-T., Choi, S., Pichler, H., & Lukin, M. D. (2020) · Physical Review X · doi:10.1103/physrevx.10.021067 · Cited by 913
Progress towards practical quantum variational algorithms
Wecker, D., Hastings, M. B., & Troyer, M. (2015) · Physical Review A · doi:10.1103/physreva.92.042303 · Cited by 716 · cited 1×
“Several QML algorithms based on PQC have been proposed [21–27], demonstrating superior performance and computational abilities compared to conventional algorithms [28–31].”
MethodQuanvolutional neural networks: powering image recognition with quantum circuits
Henderson, M., Shakya, S., Pradhan, S., & Cook, T. (2020) · Quantum Machine Intelligence · doi:10.1007/s42484-020-00012-y · Cited by 489
Cited by
Hybrid Quantum Image Classification and Federated Learning for Hepatic Steatosis Diagnosis
Lusnig, L., Sagingalieva, A., Surmach, M., Protasevich, T., Michiu, O., McLoughlin, J., Mansell, C., Petris, G. D., Bonazza, D., Zanconati, F., et al. (2024) · Diagnostics · doi:10.3390/diagnostics14050558 · Cited by 65
Roh, E. J., Baek, H., Kim, D., & Kim, J. (2024) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2024.3470328 · Cited by 35
Quantum deep learning in neuroinformatics: a systematic review
Orka, N. A., Awal, M. A., Lió, P., Pogrebna, G., Ross, A. G., & Moni, M. A. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-025-11136-7 · Cited by 35
Hayat, M. T., Allawi, Y. M., Alamro, W., Sultan, S. M., Abadleh, A., Kang, H., & Zreikat, A. I. (2025) · Diagnostics · doi:10.3390/diagnostics15212673 · Cited by 29
Comparison of machine learning algorithms for classification of Big Data sets
Singh, B., Indu, S., & Majumdar, S. (2024) · Theoretical Computer Science · doi:10.1016/j.tcs.2024.114938 · Cited by 16
Metadata
- Authors
- Ajlouni, N., Özyavaş, A., Takaoğlu, M., Takaoğlu, F., & Ajlouni, F.
- Affiliations
- TUBITAK BILGEM (TR)
- Atlas Üniversitesi (TR)
- Türkiye Bilimsel ve Teknolojik Araştırma Kurumu (TR)
- Corresponding author
- Ajlouni, Naim
- Venue
- BMC Medical Imaging
- Published
- 2023-09-14
- Volume
- 23
- Issue
- 1
- Pages
- 126-126
- ISSN
- 1471-2342 (ISSN-L)
- PMC
- PMC10500912
- OpenAlex
- W4386746261
- Cited by
- 97
- References
- 40
- Citations by year
- 2026: 53 · 2025: 29 · 2024: 15
- FWCI
- 9.12
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- BMC Medical Imaging (publishedVersion, cc-by)
- Research Square (acceptedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Quantum Computing Algorithms and Architecture
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Keywords
- quantum machine learning, medical image classification, brain MRI classification, adaptive optimization algorithm, adaptive momentum, variational quantum circuits, quantum convolutional neural networks, diagnostic imaging
- MeSH terms
- Machine Learning, Algorithms, Brain, Humans, Motion, Neural Networks, Computer
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Medical image diagnosis based on adaptive Hybrid Quantum CNN. https://app.cochat.ai/papers/10.1186/s12880-023-01084-5
- The paper
- Ajlouni, N., Özyavaş, A., Takaoğlu, M., Takaoğlu, F., & Ajlouni, F. (2023). Medical image diagnosis based on adaptive Hybrid Quantum CNN. BMC Medical Imaging. https://doi.org/10.1186/s12880-023-01084-5
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.