AICVv

docx

School

Daystar University, Machakos *

*We aren’t endorsed by this school

Course

610

Subject

Medicine

Date

Nov 24, 2024

Type

docx

Pages

Uploaded by MagistrateExplorationPelican23

Title of the proposal: Computer Visioning and Artificial Intelligence for Radiology: Challenges, Difficulties and Criteria Successful for early Diagnostics Section 2: Abstract. Radiology is a field of medicine that concentrates on the identification and treatment of maladies using visualization methods. It has grown to be a vital part of clinical practice and an integral component of modern medicine. It includes recommendations for therapy and continuing disease care, extending beyond simple disease identification. A patient's health can be visually documented, treatment can be monitored, and quick therapeutic interventions can be guided with the aid of examinations such as computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), ultrasound, and X-rays. By streamlining treatment tailoring, enhancing the results of therapy, and minimizing adverse reactions, medical imaging has a substantial effect on patient care. Medical imaging allows these intricate perspectives on physiological, molecular, and anatomical disease processes. Artificial intelligence and computer vision have become increasingly important in medicine over the past decade, especially in image technology. Artificial intelligence and computer vision are used in radiology and image modalities in medical applications. Various imaging modalities, such as radiography, ultrasound, dermoscopy, and computer tomography, offer several chances for building automated systems that aid in the diagnosis, focussing on the unique properties of AI. There is a good chance that such technologies will have a big effect on public health. General practitioners can to use these technologies to expand their capacities and standardize decision-making processes, especially in areas where a shortage of medical professionals is reportedly present. In spite of these benefits, there are always certain difficulties to be overcome, and whenever the performance data is updated across several applications, there is always space for improvement. One of the intrinsic

difficulties in integrating AI into imaging is that the literature has not yet clarified what the ultimate goal of AI techniques in imaging will be, nor how they will affect radiologists or diagnosis. Through qualitative design, the proposed study explores future developments, challenges, and difficulties that computer vision and artificial intelligence may face over the coming years. The study aims to establish success criteria to build systems to solve more complex problems that assist in early medical diagnosis. Due to the study's exploratory nature, the study proposed to adopt a qualitative approach using an integrative systematic literature review and interviewing experts, specialists, and practitioners on the use of artificial intelligence in medical practice. The qualitative content analysis will be used as a suitable method for data analysis. Section 3: Proposal Key words: artificial intelligence, computer vision, medical diagnosis, radiologists, future trends, machine learning Section 4: Research team and Summary Section 5: Budget and Timeline Section 6: RDIA focus areas: Health and wellness Section 7: Background and Motivation Background information and the current challenges : While the exact role of AI in imaging applications is unknown, it is often preferable to perform research on smaller, more uniform, and more well-defined populations. In order to properly address the patient numbers issue, the population under investigation needs to be clearly established. (Thrall et al., 2018) . AI applications typically demand significant training cases to

generate high quality, well-labelled training data sets. Institutional racism, racial discrimination, xenophobia related intolerance, along legally enforceable rights may limit access to image data among institutions. Failure to create a sufficiently large training set is a possible overfitting risk that may lead to invalid or ungeneralisable results (Cho et al., 2015) . The pitfall of overfitting was noted previously (Jyotiyana & Kesswani, 2021) . AI programs can introduce errors when applied to a set of image data that does not use the same protocol used during training. On a superficial level, if AI application in radiology was trained using the English language, it might not work but with extra training if it is given data in a different language. Similar considerations apply to the tolerance or latitude, as no information is available on how much latitude AI programs have to accommodate changes in image acquisition protocols (Dash et al., 2020). Why is it important? The biggest obstacle to the application of AI in medical imaging might be AI's intrinsic limitations in differentiating between normal and pathological biological data. Normal ranges usually indicate a specific number of standard deviations from the population mean that is supposedly average. As such, "abnormal" results for all tests and measurements will be found in a given ratio of actually normal individuals. It will be difficult to define normal and abnormal nominal criteria when, for example, defining limitations for organ sizes, will present a dilemma for researchers studying artificial intelligence. What is the problem or question you are trying to solve? One of the inherent difficulties in incorporating AI into imaging is that the literature has not yet established what the ultimate goal of AI techniques in imaging will be, nor how they impact radiologists or diagnosis. Nevertheless, the research still lacks clarity regarding the ultimate or

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

maximum use of AI techniques in imaging, their effect on radiologists, and their importance for early diagnosis. The goal of the project is to develop success criteria for systems that will be used to address more complicated issues and aid in early medical diagnosis. Literature review: What is state of the art in this area of research? In medical applications, AI and CV are utilised in radiology and image modalities (Esteva et al., 2021). Such technologies are likely to have a significant impact on public health. General practitioners will be able to leverage these technologies to increase their capabilities and standardise decision-making processes, particularly in areas where a lack of medical specialists is reported (Lee & Yoon, 2021). However, despite these advantages, some difficulties experienced, and there is room for further enhancement whenever the performance data is updated across different applications (Olveres et al., 2021). In spite of the rise of databases in the last decade, there remains a need for even more information to support the developing and implementing of reliable methods of integrating AI and CV in radiology and image modalities for medical diagnosis (Kelly et al., 2019). In the medical field, databases must be annotated by experts. Publicly available benchmark datasets can provide valuable insights into existing methodologies for comparing approaches, discovering helpful strategies, and providing guidance for developing new methodologies (Olveres et al., 2021). Some factors are responsible for the lack of updated benchmark databases. It is a common occurrence that the research data is not accessible for all, with limitations in clinical settings and Figure 1 the relation among AI, ML, DL and CV in medical imaging

a shortage of medical experts willing to provide annotations on large volumes of information (Kauppi et al., 2013). Despite years of effort assessing the efficacy of various approaches used in the medical sector, there is a notable deficiency in sufficient and well-balanced data when compared to the abundance of publicly accessible datasets in other domains, including Google's Open Images datasets. (Olveres et al., 2021). New medical workflows are needed to solve the problems described in this proposal. However, encouraging research into exploring challenges in adapting new artificial intelligence methods that rely less on big data and require less computationally (L’heureux et al., 2017). In the medical sector, cross-domain and cross-modal training and enhancement are needed, even though augmenting data and transfer learning are often used to small datasets, such as malware picture classification. (Marastoni et al., 2021). The emergence of the meta-learning paradigm is proving to be an exciting development in machine learning (Clune, 2019). Specifically, this creates a novel opportunity to deal with the limitation of the data set in medical imaging (Singh et al., 2021). Advances in CV and IA would potentially reward radiologists and biotechnology researchers with vast amounts of data, but not necessarily the ability to use these complex data effectively (Obermeyer & Emanuel, 2016) . An ultimate goal for the imaging community is to create value in patient care, particularly in the delivery of radiology services, like improving diagnostic accuracy, reducing radiologists' turnaround time, providing results to patients within a short time, and reducing the cost of care with better outcomes (Chockley & Emanuel, 2016) .

The imaging community faces many opportunities and challenges and is developing its own AI methods. These include establishing a common nomenclature, sharing image data more efficiently, and authenticating AI programs across many imaging platforms and patient populations (Ravì et al., 2016) . By using CV, AI can extract radiometric details from images in a way not obtainable by visual inspection, which can greatly enhance how image datasets can be used for diagnostic or prognostic purposes (Mei et al., 2020). An IA surveillance program could aid radiologists in prioritising their worklists if they can identify rare or positive cases for more rapid diagnosis. Through artificial intelligence, imaging will be more certain, turnaround time will be reduced, patient outcomes will be improved, and radiographers' working conditions will improve (van Leeuwen et al., 2021) . Artificial intelligence promises novel and advanced techniques for image data analysis. Using artificial intelligence in medical applications is likely to involve radiologists exploring these new treatments. In the imaging industry, AI is playing an increasingly significant role and is attracting interest from the wider community (Lee & Yoon, 2021) . The use of AI was predicted to eliminate the need for radiologists. Although the safety of AI tactics has been overblown, radiologists are likely to benefit from incorporating them into their practices (Pesapane et al., 2020) . However, currently, it is challenging to obtain technical expertise and computing power, both of which will take time (Iqbal et al., 2021) . AI's full or final role in imaging, or how it will influence or affect radiologists, is still being developed (Ahuja, 2019) . The clear implication is that AI provides a promising array of techniques for image analysis. These tools could positively influence the study of image data in the future. These tools should be explored vigorously (Raoof et al., 2021) .

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

There are circumstantial and intrinsic challenges associated with the potential for AI in medicine (Currie et al., 2019) . Circumstantial challenges depend on societal behaviour, while inherent challenges depend on technology and science. Even if none of these challenges stands in the way of AI progress in medicine, they should be addressed (Sureka, 2021) . Radiology, as a speciality, has faced as many circumstantial challenges as any other speciality in medicine. This is because there has been rapid technological change. Despite the advantages of working with digital systems, the adoption of artificial intelligence in radiology may face cultural barriers based on concerns that machines will replace radiologists (Pesapane et al., 2018) . Obermeyer and Emanuel (2016) predicted that soon, radiologists and anatomic pathologists will be displaced by machine learning for much of their work, and the accuracy of machines will soon surpass that of humans. Chockley and Emanuel (2016) also stated that an ML-driven future might lead to the demise of radiology as a thriving speciality within five to ten years. Another concern for automated medical imaging methods is the need for a strong source of truth per diagnosis in any potential artificial intelligence program, whether the learning is supervised or unsupervised (Di Noto et al., 2020) . Patients' results or the outcomes of other standard tests other than the imaging method being studied can be considered sources of truth. However, each artificial intelligence program must be specified explicitly as to the source of truth it uses (Sureka, 2021) . In addition, the current medical institutions lack conventional computing systems that provide results within a short time in a clinically pertinent time frame in an emergency or urgent diagnosis (Organization, 2020; Thrall et al., 2018) . It is necessary for practising radiologists to become familiar with artificial intelligence (Tang et al., 2018) . Still, they are not required to become experts in artificial intelligence research

and algorithm design to use AI-based results effectively (Thrall et al., 2018) . In terms of equipment and software, there are significant upfront costs associated with AI. AI programs are likely to significantly impact clinical decisions based on FDA approval (Asan et al., 2020) . However, little information is available on how AI programs will be validated or if and how individuals will be credentialed in their use. A second concern is the question of liability arising from AI programs' "black box" nature. Defining data ownership and determining who may use data will be a legal issue within institutions (Thrall et al., 2018) . RESEARCH AIM AND OBJECTIVES What are you aiming to accomplish? • To ascertain the present level of medical imaging's ability to use CV and AI. • To investigate the obstacles and difficulties biomedical researchers, developers, and professionals have while implementing artificial intelligence in medical imaging. • To learn about the latest difficulties that the scientific community is encountering in creating and putting into practice completely automated real-time clinical tasks that will aid in complex diagnosis and procedures. • To ascertain the most reliable method for creating a validating source of truth. • To investigate the possibility of creating AI programs that are resistant to protocol changes. • To ascertain whether the patient population or populations within a program are legitimate, can be utilized in AI imaging programs, and have the same level of tolerance.

How are you going to do the work? Through qualitative design, the study will explore emergent trends, challenges, and difficulties that computer vision and artificial intelligence may encounter in the near future. It is an exploratory study of the literature and opinions informed by experts, specialists and professional's practitioners of fellows and trainees of speciality colleges, such as ophthalmology, radiology/radiation oncology, and dermatology in Australian universities. Why will this approach be successful? The literature review and interviews appear to be the most appropriate methods for conducting this research. Reviewing the literature provides a holistic approach to data collection for a comprehensive overview of artificial intelligence applications that apply to some of the most crucial global medical challenges and diseases (Di Vaio et al., 2020). Section 10: Research plan and methodology JUL - 22 to APR - 23 APR - 23 to SEP - 23 SEP - 23 to FEB – 24 FEB - 24 to JUL – 24 JUL - 24 to JUL - 25 9 Month 6 M 6M 6 M One year

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

Research training Proposal development Literature review to help refine the primary research question Preparing for data gathering Preparing the interview protocol Conducting Preparing all fieldwork needed confirmation seminar Applying for ethics (if needed) Completing the full literature and research study design; A pilot search in the WOS, Scopus and PubMed databases Starting Gathering the data Recruiting study participants Collecting the data, including: Conducting interviews Starting the data analysis Drafting the 1 st Publication Mid-Seminar Completion of data analysis Drafting the 2nd Publication Complete Writing up Drafting the 3rd Publication Supplementary literature review Writing up Submission Publications and Submission (Publications, Preparing Manuscript, Reviewing drafts and Binding) Section 11: Management plan and timeline (Provide high level overview of the management plan of the project.)

Section 12: Resource allocation Section 13: Success and impact The ultimate goal of the proposed study is to explore the current best practice of AI approaches in imaging applications. The study stands to investigate the challenges and difficulties of acclimating AI in medical imaging, particularly radiology, for early diagnostics. The findings are fundamental to develop a practical framework as a criteria guideline. Responsibilities

References Ahuja, A. S. (2019). The impact of artificial intelligence in medicine on the future role of the physician. PeerJ , 7 , e7702. Asan, O., Bayrak, A. E., & Choudhury, A. (2020). Artificial intelligence and human trust in healthcare: focus on clinicians. Journal of medical internet research , 22 (6), e15154. Azungah, T. (2018). Qualitative research: deductive and inductive approaches to data analysis. Qualitative research journal . Brownlee, J. (2019). Deep Learning for Computer Vision: Image Classification, Object Detection, and Face Recognition in Python . Machine Learning Mastery. https://books.google.com.sa/books?id=DOamDwAAQBAJ Cho, J., Lee, K., Shin, E., Choy, G., & Do, S. (2015). How much data is needed to train a medical image deep learning system to achieve necessary high accuracy? arXiv preprint arXiv:1511.06348 . Chockley, K., & Emanuel, E. (2016). The end of radiology? Three threats to the future practice of radiology. Journal of the American College of Radiology , 13 (12), 1415-1420. Clune, J. (2019). AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence. arXiv preprint arXiv:1905.10985 . Currie, G., Hawk, K. E., Rohren, E., Vial, A., & Klein, R. (2019). Machine learning and deep learning in medical imaging: intelligent imaging. Journal of medical imaging and radiation sciences , 50 (4), 477-487. Dash, S., Acharya, B. R., Mittal, M., Abraham, A., & Kelemen, A. (2020). Deep learning techniques for biomedical and health informatics . Springer.

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

Di Noto, T., Mannil, M., Aerts, H., & Kadian, C. (2020). Artificial Intelligence and Radiomics: Outlook into the Future. In Neuroimaging Techniques in Clinical Practice (pp. 335-342). Springer. Di Vaio, A., Palladino, R., Hassan, R., & Escobar, O. (2020). Artificial intelligence and business models in the sustainable development goals perspective: A systematic literature review. Journal of Business Research , 121 , 283-314. Esteva, A., Chou, K., Yeung, S., Naik, N., Madani, A., Mottaghi, A., Liu, Y., Topol, E., Dean, J., & Socher, R. (2021). Deep learning-enabled medical computer vision. NPJ digital medicine , 4 (1), 1-9. Flick, U. (2015). Introducing research methodology: A beginner's guide to doing a research project . Sage. Fourcade, A., & Khonsari, R. (2019). Deep learning in medical image analysis: A third eye for doctors. Journal of stomatology, oral and maxillofacial surgery , 120 (4), 279-288. Hamet, P., & Tremblay, J. (2017). Artificial intelligence in medicine. Metabolism , 69 , S36-S40. Hidalgo, C. A., Orghian, D., Canals, J. A., De Almeida, F., & Martin, N. (2021). How humans judge machines . MIT Press. Hoepfl, M. C. (1997). Choosing qualitative research: A primer for technology education researchers. Volume 9 Issue 1 (fall 1997) . Hopia, H., Latvala, E., & Liimatainen, L. (2016). Reviewing the methodology of an integrative review. Scandinavian journal of caring sciences , 30 (4), 662-669. Iqbal, M. J., Javed, Z., Sadia, H., Qureshi, I. A., Irshad, A., Ahmed, R., Malik, K., Raza, S., Abbas, A., & Pezzani, R. (2021). Clinical applications of artificial intelligence and

machine learning in cancer diagnosis: Looking into the future. Cancer cell international , 21 (1), 1-11. Jyotiyana, M., & Kesswani, N. (2021). Introduction to Deep Learning in Health Informatics. Biomedical Data Mining for Information Retrieval: Methodologies, Techniques and Applications , 237-261. Kauppi, T., Kämäräinen, J.-K., Lensu, L., Kalesnykiene, V., Sorri, I., Uusitalo, H., & Kälviäinen, H. (2013). Constructing benchmark databases and protocols for medical image analysis: Diabetic retinopathy. Computational and mathematical methods in medicine , 2013 . Kelly, C. J., Karthikesalingam, A., Suleyman, M., Corrado, G., & King, D. (2019). Key challenges for delivering clinical impact with artificial intelligence. BMC Medicine , 17 (1), 195. https://doi.org/10.1186/s12916-019-1426-2 L’heureux, A., Grolinger, K., Elyamany, H. F., & Capretz, M. A. (2017). Machine learning with big data: Challenges and approaches. IEEE Access , 5 , 7776-7797. Lee, D., & Yoon, S. N. (2021). Application of artificial intelligence-based technologies in the healthcare industry: Opportunities and challenges. International journal of environmental research and public health , 18 (1), 271. Lyall, C., & King, E. (2016). Using qualitative research methods in biomedical innovation: the case of cultured red blood cells for transfusion. BMC research notes , 9 , 267-267. https://doi.org/10.1186/s13104-016-2077-4 Marastoni, N., Giacobazzi, R., & Dalla Preda, M. (2021). Data augmentation and transfer learning to classify malware images in a deep learning context. Journal of Computer Virology and Hacking Techniques , 17 (4), 279-297.

Maxwell, J. A. (2008). Designing a qualitative study. The SAGE handbook of applied social research methods , 2 , 214-253. Nilsson, N. J. (2009). The quest for artificial intelligence . Cambridge University Press. Obermeyer, Z., & Emanuel, E. J. (2016). Predicting the future—big data, machine learning, and clinical medicine. The New England journal of medicine , 375 (13), 1216. Olveres, J., González, G., Torres, F., Moreno-Tagle, J., Carbajal-Degante, E., Valencia- Rodríguez, A., Méndez-Sánchez, N., & Escalante-Ramírez, B. (2021). What is new in computer vision and artificial intelligence in medical image analysis applications. Quantitative Imaging in Medicine and Surgery , 11 , 3830-3853. https://doi.org/10.21037/qims-20-1151 Organization, W. H. (2020). Outbreak preparedness and resilience. Pesapane, F., Tantrige, P., Patella, F., Biondetti, P., Nicosia, L., Ianniello, A., Rossi, U. G., Carrafiello, G., & Ierardi, A. M. (2020). Myths and facts about artificial intelligence: why machine-and deep-learning will not replace interventional radiologists. Medical Oncology , 37 (5), 1-9. Pesapane, F., Volonté, C., Codari, M., & Sardanelli, F. (2018). Artificial intelligence as a medical device in radiology: ethical and regulatory issues in Europe and the United States. Insights into imaging , 9 (5), 745-753. Raoof, S. S., Jabbar, M., & Tiwari, S. (2021). Foundations of deep learning and its applications to health informatics. Deep Learning in Biomedical and Health Informatics: Current Applications and Possibilities , 1.

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

Ravì, D., Wong, C., Deligianni, F., Berthelot, M., Andreu-Perez, J., Lo, B., & Yang, G.-Z. (2016). Deep learning for health informatics. IEEE journal of biomedical and health informatics , 21 (1), 4-21. Ryan, F., Coughlan, M., & Cronin, P. (2009). Interviewing in qualitative research: The one-to- one interview. International Journal of Therapy and Rehabilitation , 16 (6), 309-314. Sarma, P. (2022). Technology Focus: Digital Data Acquisition (January 2022). Journal of Petroleum Technology , 74 (01), 90-91. Singh, R., Bharti, V., Purohit, V., Kumar, A., Singh, A. K., & Singh, S. K. (2021). MetaMed: Few-shot medical image classification using gradient-based meta-learning. Pattern Recognition , 120 , 108111. Sureka, C. (2021). Artificial Intelligence in Medical Imaging. In Artificial Intelligence Theory, Models, and Applications (pp. 47-74). Auerbach Publications. Tang, A., Tam, R., Cadrin-Chênevert, A., Guest, W., Chong, J., Barfett, J., Chepelev, L., Cairns, R., Mitchell, J. R., & Cicero, M. D. (2018). Canadian Association of Radiologists white paper on artificial intelligence in radiology. Canadian Association of Radiologists Journal , 69 (2), 120-135. Thrall, J. H., Li, X., Li, Q., Cruz, C., Do, S., Dreyer, K., & Brink, J. (2018). Artificial intelligence and machine learning in radiology: opportunities, challenges, pitfalls, and criteria for success. Journal of the American College of Radiology , 15 (3), 504-508. Torraco, R. J. (2016). Writing integrative literature reviews: Using the past and present to explore the future. Human Resource Development Review , 15 (4), 404-428.

van Leeuwen, K. G., de Rooij, M., Schalekamp, S., van Ginneken, B., & Rutten, M. J. C. M. (2021). How does artificial intelligence in radiology improve efficiency and health outcomes? Pediatric Radiology . https://doi.org/10.1007/s00247-021-05114-8 Wang, J., Zhu, H., Wang, S.-H., & Zhang, Y.-D. (2021). A review of deep learning on medical image analysis. Mobile Networks and Applications , 26 (1), 351-380. Whittemore, R., & Knafl, K. (2005). The integrative review: updated methodology. Journal of advanced nursing , 52 (5), 546-553.

AICVv

Related Documents