Content uploaded by Kun Hing Yong
Author content
All content in this area was uploaded by Kun Hing Yong on Mar 31, 2024
Content may be subject to copyright.
JCMC/ Vol 14/ No. 1/ Issue 47/ Jan-Mar, 2024
ISSN 2091-2889 (Online) ISSN 2091-2412 (Print)
ABSTRACT
Breast cancer is a signicant global health challenge that aects both men and women, leading
to cause-specic deaths. Current early screening intervenons, such as digital mammography
(DM), are suscepble to high false-posives and false-negaves. This paper explores the potenal
of convoluonal neural network (CNN), a form of arcial intelligence (AI), to support screening
mammography with the aim to enhance accuracy in lesion detecon, image classicaon and
diagnosc predicon. Because the adopon of AI in cancer diagnosis is sll in its infancy, the
objecve of this paper is to provide insight into the benets and limitaons of deep learning-
based approaches to detect and diagnose cancer. An analysis of the implementaon of CNN in
AI-screening mammography models was conducted, using the SWOT strategic analysis tool.
Internal strengths that improve the predicve accuracy of CNN include transfer learning and
data augmentaon, whereas the internal weaknesses include a lack of data standardisaon and
reproducibility. External opportunies consist of increased sensivity in dierenang between
microcalcicaons and non-tumorous structures, improved predicve diagnosis and reduced
workload. Nevertheless, integraon within clinical sengs must also consider the external threats
of breaching paent privacy, automaon biases and the role of clinical judgement.
Journal of Chitwan Medical College 2024;14(47):89-94
Available online at: www.jcmc.com.np
REVIEW ARTICLE
A SWOT ANALYSIS OF BREAST CANCER DIAGNOSIS IN DIGITAL MAMMOGRAPHY USING DEEP
CONVOLUTIONAL NEURAL NETWORK
Elizabeth Yong1, Yen Nee Teo2, Leanne McKnoulty3, Ajeevan Gautam4, Rajib Chaulagain5, Kun Hing Yong4,*
1Indooroopilly State High School, Brisbane, Queensland 4068, Australia
2Instute of Malaysian and Internaonal Studies, Naonal University of Malaysia, Bangi 43600, Selangor, Malaysia
3GUPSA editor in residence, Grith University, Nathan, Queensland 4111, Australia
4School of Medicine and Denstry, Grith University, Nathan, Queensland 4111, Australia
5Department of Oral Pathology, Chitwan Medical College, Bharatpur, Chitwan 44207, Nepal
ISSN 2091-2889 (Online) ISSN 2091-2412 (Print)
E
S
T
D
2
0
1
0
J
O
U
R
N
A
L
O
F
C
H
I
T
W
A
N
M
E
D
I
C
A
L
C
O
L
L
E
G
E
Received: 24 Jan, 2024
Accepted: 11 Mar, 2024
Published: 30 Mar, 2024
Key words: Arcial intelligence; Breast
cancer; Convoluonal neural network; Digital
mammography; SWOT analysis.
*Correspondence to: Kun Hing Yong, School of
Medicine and Denstry, Grith University, 170 Kessels
Rd, Nathan Queensland 4111, Australia.
Email:
DOI:
Yong E, Teo YN, McKnoulty L, Gautam A, Chaulagain
R, Yong KH. HA SWOT analysis of breast cancer
diagnosis in digital mammography using deep
convoluonal neural network. Journal of Chitwan
Medic al Colle ge.2024;14(47):89-94.
JCMC
INTRODUCTION
Breast cancer is presently the most commonly diagnosed cancer
in women globally, and the second-leading cause of mortality
from cancer. Accurate cancer diagnosis of symptomac
paents at an early stage is pernent to improve cancer
outcomes, thereby reducing cause-specic deaths. In the
early 1990s in Australia, screening mammography programs
were implemented for the early detecon and treatment of
breast cancer, but their accuracy in sensivity and specicity
remains error-prone, leading to the reporng of false-posives
and false-negaves, respecvely. The addional imaging tests
and biopsies ensuing false-posive recalls can contribute to
unnecessary emoonal stress for the paent. Similarly, health
hazards from high radiaon exposure should also be considered.
Errors in interpretaon and detecon of abnormalies can
be aributed to dierent breast densies, small tumours or
arfacts.1 Another limitaon is subjecvity in image analysis
due to varied percepons across interpreters, known as inter-
reader variability. During double reading to improve diagnosc
accuracy, two radiologists independently read the same
screening mammography.2 Despite image analysis performed
manually by experts, factors such as fague and decreased
aenon can adversely aect the results ndings. Furthermore,
double reading is labour intensive, implying that the me
constraints on clinical evaluaons and examinaons can lead
to a delegaon of tasks from radiologists to other physicians
or breast clinicians. This can lead to unfavourable outcomes
for the paents, being subjected to higher posive recall rates
and false-posive interpretaons, because physicians may lack
in sucient radiological knowledge to exert accurate clinical
judgement.3
With rapid development in compung power and data, AI has
been increasingly integrated in clinical sengs. Among them
is machine learning and, in parcular, deep learning with
CNN diagnosc-based approaches, whereby the technology is
trained to recognise complex paerns from raw input with its
mul-layered networks and make accurate connecons based
on the context. Its ulity in lesion detecon, image classicaon
and diagnosc predicon enable addional aid to radiologists
to achieve higher accuracy when interpreng DM, thereby
serving as a prospecve applicaon to improve diagnosis
of breast cancer.4 The applicaons of these technological
innovaons have understandably raised concerns among
healthcare professionals, in regard to its feasibility and
JCMC/ Vol 14/ No. 1/ Issue 47/ Jan-Mar, 2024
ISSN 2091-2889 (Online) ISSN 2091-2412 (Print)
diagnosc ecacy. To address the concerns of AI applicaons
in medical imaging, an understanding of the benets and
limitaons of AI tools is necessary.5
METHODS
A literature review of research published during the last 5
years was conducted to evaluate the strengths, weaknesses,
opportunies and threats (SWOT) of CNN in AI models used
to diagnose breast cancer. A brief analysis is provided, while
the primary points are outlined in Table 1. This SWOT analysis
forms the basis for governmental decision-makers and health
care providers to understand the potenal implementaon of
AI within clinical sengs, and to consider future improvements
in approaching the problem.
Secon 2 introduces the funconality of CNN, Secon 3
elucidates the strengths of applying CNN in mammography
to diagnose breast cancer, Secon 4 explains the external
opportunies, and Secons 5 and 6 discuss the weaknesses
and threats or ethical challenges. Finally, Secon 7 presents
the suggested future direcons and the conclusion.
In deep learning, a CNN is a class of deep neural network that
uses algorithms to process a large quanty of data with a grid
paern, notably in image-related analysis.6 CNN is employed
for image examinaon, idencaon or classicaon because
it can eciently extract features from images and simplify
them for beer analysis. It consists of three disnct layers
with funcons that interconnect each other, namely an input
layer, mulple hidden layers and an output layer. The inial
DM image undergoes ltering in the rst convoluonal layer,
which enhances the features, removes unwanted noise, and
helps to dierenate the edges and shapes of the region
under invesgaon. Subsequent convoluonal layers enhance
the feature paerns to facilitate idencaon of tumour
contour and enable the extracon of specic features, such as
structural paerns or dominant outliers in the image, making
CNN highly ecient for image processing.7 The pooling layer
lters the minimum, maximum, mean or median of the set of
pixels within the image that fall within the lter, to reduce the
spaal size and maintain only the most crucial informaon.8
Decreasing the parameters increases the processing speed.
The informaon is subsequently passed through the fully
connected layer, where extracon of inputs from feature
analysis and applicaon of weights and predicts the output
into classes of cancer. For example, in the study by Ragab and
colleagues9, the fully connected layer classied abnormal areas
as benign or malignant, while various other studies classied
regions as benign, malignant or without tumour. Figure 1
depicts the structure of a classic CNN architecture.
DISCUSSION
Transfer learning
Transfer learning refers to leveraging the learned features of
a pre-trained model as the foundaon for training a model to
perform a new task. It takes advantage of the fact that neural
networks trained on large databases of images, such as those
with ImageNet, have learned and established parameters in
the early layers relevant to numerous visual tasks, despite the
specic task they are programmed to perform.10 Salehi and
colleagues explained that certain funcons of CNNs in lower
layers, such as those dedicated for edge, texture and paern
detecons, can be calibrated and applied to higher layers of
the network.10 However, the specic features that must be
learned will increase in complexity where, for instance, the
output layer would only respond to images of a specic tumour
that it had been trained to detect. Thus, using a pre-trained
model and customising the new model with addional new
layers and adjustments to the number of neurons or classes
depending on the specic task requirements has the benet of
minimising training me and requires limited data. This means
earlier models can be rened and adapted to various tasks,
including detecng and classifying lesions, without retraining
a deep neural network from scratch.
pooling layer, and fully connected layer
Note. The nal output is classied as normal, benign or
malignant.
In medical imaging where the number of fully annotated
mammograms available is limited, training a deep learning
model with data augmentaon ensures improvement to the
models while also minimising data overng. Overng is
a stascal error whereby the model ts too closely to the
trained dataset and cannot be generalised to new data.11
Data augmentaon enables arcial expansion on exisng
datasets to generate modied copies and, hence, introduces
a vast variety of paerns that the model can recognise and
learn from. Improvement to data variability is demonstrated
to enhance the predicve accuracy of the AI models in
detecng suspicious regions of interest when presented with
normal and abnormal DMs.12,13 This provides the radiologist
with psychological support, by reducing the cognive burden
associated with idenfying potenal lesion regions.
For example, GAN-based augmentaon, an unsupervised deep
learning method that extracts hidden properes from data to
formulate its decision-making process, has shown potenal
JCMC/ Vol 14/ No. 1/ Issue 47/ Jan-Mar, 2024
ISSN 2091-2889 (Online) ISSN 2091-2412 (Print)
to improve accuracy in mass classicaon aer geometric
transformaons from unrelated masses or increase in noise
distorons.12 As such, it has also been a widely used approach
in breast mass detecon and mass segmentaon.13 As the
use of data augmentaon methods expands, it is pernent to
evaluate the quality of the output and recognise that building
upon minimal databases can restrict the generalisaon
ability of the model and potenally reinforce inherent biases.
With higher resoluon DM images, convenonal computer-
aided diagnosis (CAD) models can disnguish between benign
and malignant lesions by assessing their greyscale levels,
homogeneity, gradient, paerns and shape.14 However,
because dense breast ssue appears white and has similar
shade and intensity values as tumorous regions containing
microcalcicaons, dense breast ssue, with relavely high
amounts of glandular ssue and brous connecve ssue,
can hide lesions and is prone to misdiagnosis and reporng of
false negaves. With AI screening, it can perform detecon of
potenally tumorous region and compare its intensity value
with other regions of the breast followed by segmentaon
of the tumour area surrounded with malignant ssues.15
This can reduce the lower sensivity from human perceptual
error, because it separates pixels of cancer region from normal
region. Geras et al. showed that the addion of the deep
learning method, which learns the intermediate and abstract
representaons of the data, can improve accuracy in lesion
classicaon in DMs, reaching similar sensivity to radiologists’
assessment.14
Given the large processing capacity of AI, its capability of
analysing and processing data from wide-ranging sources,
including medical images, laboratory test results and paent
history, enables idencaon of paerns and abnormalies
that may otherwise be missed by human experts. Missed
microcalcicaons can be aributed to their small size or
concealment by overlying high amounts of brous and glandular
ssues.16 Therefore, implemenng AI in mammography has the
potenal to increase sensivity in dierenang between the
microcalcicaons and non-tumorous anatomic structures,
such as increased breast density. It employs image processing
techniques to spaally lter the DM and improve signal-
to-noise rao, yielding higher sensivity for detecng true
abnormalies.15 In a study by Kim et al., the classicaon
performance of AI-CAD demonstrated a higher accuracy value
of 0.938–0.970 compared to an accuracy value of 0.810–0.881
achieved by radiologists.17 Findings by Liu and colleagues
also reported that combining the deep learning model into
mammography aained similar diagnosc performance to
that of an experienced radiologist, and signicantly surpassed
the performance of a junior radiologist (p=0.029; p<0.05).18
The improvement indicated promising results in reducing the
quanty of unnecessary biopsies performed, showing potenal
for early detecon and intervenon of breast cancer.
Reduced workload
Numerous European countries have employed double reading
with arbitraon, whereas the United States typically has
employed single reading with CAD.19 While standard double
reading has been shown to reduce recall rates, it is labour
intensive. A study by Dembrower and colleagues compared
the cancer detecon rates and eciency of varying methods
of interpretaon: single reading by AI, double reading by two
radiologists, double reading by one radiologist and AI and triple
reading by two radiologists and AI.20 The ndings suggested
that the performance for triple reading (95% CI 1.04–1.11)
outperformed the double reading by one radiologist and AI or
by two radiologists (95% CI 1.00–1.09). Triple reading increased
recalls by 5% and consensus discussion by 50%, while double
reading by one radiologist and AI decreased recalls by 4%
with a reasonable number of consensus discussion. In triple
reading, the percepon of the combined radiologists was
favoured over the percepon of the AI, indicang that the
ability of AI in detecng cancer was under-esmated rather
than over-esmated, explaining the slightly higher recall
rates. Because the higher abnormal interpretaon rate for AI
and one radiologist did not translate into an increased recall
rate, it would help reduce workload me, which had been
demonstrated to be by nearly 40%.20,21 Replacement of the
second reader with AI would substanally reduce the me
radiologists spend reading mammograms. Another study
by Lång and colleagues determined that mammography
screening supported with AI yielded similar cancer detecon
rate as standard double reading, with the recall rate being 0.2%
higher at 2.2% , suggesng that the use of AI in mammography
can be considered.22
Weaknesses
Standardisaon within a clinical seng can help improve
interoperability and vast exchange of health data and
informaon. This is pernent to improve performance of
the models in imaging acquision and processing, because
the quality of image acquision aects radiomic feature
calculaons, radiomics being the extensive image-based
phenotyping of abnormalies through extracon of diverse
feature values from medical images.23 Currently, insucient
standardizaon is evident in the collecon and storage of
unstructured data, as well as in the process of unifying data
that represents a single healthcare system.24 Substanal
informaon technology and systems resources is required
to implement this, and the feasibility remains under acve
invesgaon.
One method proposes using paent-reported outcomes (PROs)
and validated quesonnaires, as they are valuable survival
indicators that can benet cancer care delivery, research
and clinical operaons.25 Nonetheless, several limitaons are
JCMC/ Vol 14/ No. 1/ Issue 47/ Jan-Mar, 2024
ISSN 2091-2889 (Online) ISSN 2091-2412 (Print)
Strengths
Transfer learning:
- Minimises training me and requires limited data through modifying a pre-trained model, tailored
to suit specic requirements
Data augmentaon:
- Minimises data overng
- Improves generalisability, image recognion, segmentaon accuracy and analysis
- Enhances predicve accuracy in tumour classicaon
Weaknesses
- Lack of standardisaon limits interoperability
- Limitaons in obtaining and implemenng paent-reported outcomes
- Data reproducibility is subjected to data dri
- Lack of high-quality and mul-instuonal datasets may reduce generalisability
- Dierent mechanisms performed at each CNN layer require varying levels of complexity during
programming
Opportunies
Pixel-level image classicaon:
- Reduces false posives due to ability to discern between tumorous and non-tumorous regions
- Improves cancer detecon rates
Improved paent value through predicve diagnosis:
- Increases sensivity for detecon of true abnormalies
- Improves tumour classicaon accuracy
- Potenal to reduce quanty of unnecessary biopsies and medical costs.
Opportunies for healthcare professionals:
- Enhances reading eciency by reducing number of tests requiring radiologist interpretaon
- Reduces workload
Threats
Paent privacy:
- Breach of health and personal informaon
- Lack of transparency
Algorithmic biases:
- Biases in input training data can produce skewed results and exacerbate health care inequality
- Ethical concerns regarding the role of AI in clinical judgement
Role of human judgement:
- Medico-legal responsibility for healthcare providers if incorrect diagnosis is made
- Potenal discordance between clinical pracces and AI suggesons
- Impairment in clinical judgement from over-reliance on AI technology, resulng in potenal pa-
ent injury
- Jeopardizaon of the learning process and clinical reasoning abilies of medical students or novice
radiologists
present. These include paent-level barriers such as disability,
challenges in reading and responding to the quesonnaires or
with recalling their symptoms, clinical-level obstacles like lack
of sta training with interpreng and implemenng PROs into
clinical pracces, and service-level challenges like lack of PRO
data logging into electronic medical records within a hospital
seng.26
Data Reproducibility
Data reproducibility is limited when transferred across
healthcare systems and global communies, but even within
the training environment, data dri over me for AI algorithms
and advanced CDSS can aect their performance. This is a
result of variaons in distribuon, formang or quality of
data, awed data transformaon, absence of natural dri
when training the model or covariate shi.27 Thus, standards
must be incorporated to connuously monitor AI algorithms
and ensure their validity even if AI were to be successfully
implemented as a technological pracce in medicine due to
their evolving nature.
Threats regarding Ethical Challenges
Precision medical technology relies on extensive medical
informaon for cancer diagnosis, screening, data processing,
opmising care delivery and conducng clinical operaons.
To train models eecvely, medical researchers need access
to paents’ personal health records. However, concerns
arise regarding the potenal misuse of data, leading to issues
like identy the, insurance fraud and illegal acquision of
prescripon drugs. To ensure ethical use of paent data in
clinical pracce, medical researchers must be transparent about
how data will be used. Addionally, they should implement
robust safety measures to safeguard paent privacy and obtain
informed consent from individuals contribung their data.
Algorithmic Biases
JCMC/ Vol 14/ No. 1/ Issue 47/ Jan-Mar, 2024
ISSN 2091-2889 (Online) ISSN 2091-2412 (Print)
REFERENCES:
1. Nori J, Gill MK, Vignoli C, Bicchierai G, De Benedeo D, Di Naro F, Vanzi E,
Boeri C, Miele V. Artefacts in contrast enhanced digital mammography:
how can they aect diagnosc image quality and confuse clinical
diagnosis? Insights into Imaging. 2020;11(1):16.[DOI]
2. Salim M, Dembrower K, Eklund M, Lindholm P, Strand F. Range of
Radiologist Performance in a Populaon-based Screening Cohort of 1
Million Digital Mammography Examinaons. Radiology. 2020;297(1):33-
9. [DOI]
3. Chen Y, James JJ, Michalopoulou E, Darker IT, Jenkins J. Performance
of Radiologists and Radiographers in Double Reading Mammograms:
The UK Naonal Health Service Breast Screening Program. Radiology.
2023;306(1):102-9. [DOI]
4. Do S, Song KD, Chung JW. Basics of Deep Learning: A Radiologist’s Guide
to Understanding Published Radiology Arcles on Deep Learning. Korean
J Radiol. 2020;21(1):33-41. Epub 2020/01/11. [DOI]
5. Teo YN, Yong KH, Gautam A, Chaulagain R. Guarding our future: Harnessing
arcial intelligence to combat anmicrobial resistance and raise public
awareness. Journal of Chitwan Medical College. 2023;13(3):1-2. [DOI]
6. Nasser M, Yusof UK. Deep Learning Based Methods for Breast Cancer
Diagnosis: A Systemac Review and Future Direcon. Diagnoscs.
2023;13(1):161. [DOI]
7. Albalawi U, Manimurugan S, Varatharajan R. Classicaon of breast cancer
mammogram images using convoluon neural network. Concurrency
and Computaon: Pracce and Experience. 2022;34(13):e5803. [DOI]
8. Zafar A, Aamir M, Mohd Nawi N, Arshad A, Riaz S, Alruban A, Dua AK,
Almotairi S. A Comparison of Pooling Methods for Convoluonal Neural
Networks. Applied Sciences. 2022;12(17):8643. [DOI]
9. Ragab DA, Sharkas M, Marshall S, Ren J. Breast cancer detecon using
deep convoluonal neural networks and support vector machines. PeerJ.
2019;7:e6201. Epub 2019/02/05. [DOI]
10. Salehi AW, Khan S, Gupta G, Alabduallah BI, Almjally A, Alsolai H, Siddiqui
T, Mellit A. A Study of CNN and Transfer Learning in Medical Imaging:
Advantages, Challenges, Future Scope. Sustainability. 2023;15(7):5930.
[DOI]
11. Ying X. An Overview of Overng and its Soluons. Journal of Physics:
Conference Series. 2019;1168(2):022022. [DOI]
12. Oza P, Sharma P, Patel S, Adedoyin F, Bruno A. Image Augmentaon
Techniques for Mammogram Analysis. Journal of Imaging. 2022;8(5):141.
[DOI]
13. Desai SD, Giraddi S, Verma N, Gupta P, Ramya S, editors. Breast Cancer
Detecon Using GAN for Limited Labeled Dataset. 2020 12th Internaonal
Conference on Computaonal Intelligence and Communicaon Networks
(CICN); 2020 25-26 Sept. 2020. [DOI]
14. Geras KJ, Mann RM, Moy L. Arcial Intelligence for Mammography and
Digital Breast Tomosynthesis: Current Concepts and Future Perspecves.
Radiology. 2019;293(2):246-59. Epub 2019/09/25. [DOI]
15. Shen L, Margolies LR, Rothstein JH, Fluder E, McBride R, Sieh W.
Deep Learning to Improve Breast Cancer Detecon on Screening
Mammography. Sci Rep. 2019;9(1):12495. Epub 2019/08/31. [DOI]
16. Kressin NR, Wormwood JB, Baaglia TA, Maschke AD, Slanetz PJ, Pankowska
M, Gunn CM. Women’s Understandings and Misunderstandings of Breast
Density and Related Concepts: A Mixed Methods Study. J Womens Health
(Larchmt). 2022;31(7):983-90. Epub 2022/03/02. [DOI]
17. Kim H-E, Kim HH, Han B-K, Kim KH, Han K, Nam H, Lee EH, Kim E-K.
Changes in cancer detecon and false-posive recall in mammography
using arcial intelligence: a retrospecve, mulreader study. The Lancet
Digital Health. 2020;2(3):e138-e48. [DOI]
18. Liu H, Chen Y, Zhang Y, Wang L, Luo R, Wu H, Wu C, Zhang H, Tan W,
Yin H, Wang D. A deep learning model integrang mammography
and clinical factors facilitates the malignancy predicon of BI-RADS 4
microcalcicaons in breast cancer screening. European Radiology.
2021;31(8):5902-12. [DOI]
CONCLUSION
Current evidence regarding the integraon of AI in clinical
sengs has shown promising results in that AI-supported
screening mammography improves cancer detecon rates or
is level with senior radiologists, while also enhancing paent
outcomes and alleviang radiologists’ workload. A main
advantage is its enhanced sensivity in discerning between
benign and malignant lesions from dense breast ssues,
a challenging diagnosis, thereby minimising perceptual
errors. This can improve accuracy in diagnosc performance
and facilitate predicve diagnosis for early intervenon.
Nevertheless, the availability of well-curated datasets to
ensure high-quality result outcomes by AI systems that enable
sucient, reliable data generalisaon and cancer detecon
is yet to be assured. As the results of this paper showcase,
considerable risks could emerge that impact the accuracy of the
data and, if not migated, would aect the paent safety. These
incorporate ethical issues around medical responsibility for any
diagnosc errors made, human oversight and transparency.
Thus, investment to support clinical trials in researching and
evaluang the outcomes and performance of AI algorithms
on the paent and providers regarding breast screening
mammography is encouraged, to validate the ecacy, validity
and reliability when applied as roune clinical pracce.
Bias within AI algorithms is aected by the bias within the data
they are trained on. If a dataset is biased towards a parcular
demographic group, the validity in the AI-generated results
to predict the cancer outcomes of individuals from other
demographic groups is reduced – either over-represenng
or under-represenng certain populaons. To prevent
perpetuaon of inequalies in healthcare by AI algorithms
that may contribute to potenal harm, diverse and more
representave range of datasets should be used instead, while
inherent biases should undergo careful invesgaon to ensure
they are not overlooked.
Role of Human Judgement
Although radiologists are blinded to the output of the AI system
to prevent double reading or over-reliance on AI, the validity of
the consensus decision may be inuenced depending on the
under- or over-esmaon of the accuracy of AI systems.19 This
will result in variaons in recall rates and cancer detecon. A
strength may be a reducon in recall rates by introducing higher
specicity by experts to migate the higher cancer detecon
rates of AI. However, over-reliance on AI could lead clinicians to
overlook their crical clinical judgement, irrespecve of their
experience. As such, it can increase the risks of accountability
when performing incorrect diagnosis, as recommended by AI,
which results in avoidable harm to paents.
JCMC/ Vol 14/ No. 1/ Issue 47/ Jan-Mar, 2024
ISSN 2091-2889 (Online) ISSN 2091-2412 (Print)
19. Taylor-Phillips S, Snton C. Double reading in breast cancer screening:
consideraons for policy-making. Br J Radiol. 2020;93(1106):20190610.
Epub 2019/10/17. [DOI]
20. Dembrower K, Crippa A, Colón E, Eklund M, Strand F. Arcial intelligence
for breast cancer detecon in screening mammography in Sweden: a
prospecve, populaon-based, paired-reader, non-inferiority study. The
Lancet Digital Health. 2023;5(10):e703-e11. [DOI]
21. Rodriguez-Ruiz A, Lång K, Gubern-Merida A, Teuwen J, Broeders M,
Gennaro G, et al. Can we reduce the workload of mammographic
screening by automac idencaon of normal exams with arcial
intelligence? A feasibility study. Eur Radiol. 2019;29(9):4825-32. Epub
2019/04/18. [DOI]
22. Lång K, Josefsson V, Larsson A-M, Larsson S, Högberg C, Sartor H, et al.
Arcial intelligence-supported screen reading versus standard double
reading in the Mammography Screening with Arcial Intelligence trial
(MASAI): a clinical safety analysis of a randomised, controlled, non-
inferiority, single-blinded, screening accuracy study. The Lancet Oncology.
2023;24(8):936-44. [DOI]
23. Li XT, Huang RY. Standardizaon of imaging methods for machine learning
in neuro-oncology. Neurooncol Adv. 2020;2(Suppl 4):iv49-iv55. Epub
2021/02/02. [DOI]
24. Sedlakova J, Daniore P, Horn Wintsch A, Wolf M, Stanikic M, Haag
C, Sieber C, Schneider G, Staub K, Alois Elin D, Grübner O, Rinaldi F,
von Wyl V. Challenges and best pracces for digital unstructured data
enrichment in health research: A systemac narrave review. PLOS Digit
Health. 2023;2(10):e0000347. Epub 2023/10/11. [DOI]
25. Camini C, Magliea G, Dioda F, Puntoni M, Marcomini B, Lazzarelli S,
Pinto C, Perrone F. The Eects of Paent-Reported Outcome Screening
on the Survival of People with Cancer: A Systemac Review and Meta-
Analysis. Cancers (Basel). 2022;14(21). Epub 2022/11/12. [DOI]
26. Nguyen H, Butow P, Dhillon H, Sundaresan P. A review of the barriers
to using Paent-Reported Outcomes (PROs) and Paent-Reported
Outcome Measures (PROMs) in roune cancer care. J Med Radiat Sci.
2021;68(2):186-95. Epub 2020/08/21. [DOI]
27. Jacob T. Shreve M, Sadia A. Khanani M, Tua C. Haddad M. Arcial
Intelligence in Oncology: Current Capabilies, Future Opportunies, and
Ethical Consideraons. American Society of Clinical Oncology Educaonal
Book. 2022(42):842-51. [DOI]