An increasing array of tools is being developed using artificial intelligence (AI) and machine learning (ML) for cancer imaging. The development of an optimal tool requires multidisciplinary engagement to ensure that the appropriate use case is met, as well as to undertake robust development and testing prior to its adoption into healthcare systems. This multidisciplinary review highlights key developments in the field. We discuss the challenges and opportunities of AI and ML in cancer imaging; considerations for the development of algorithms into tools that can be widely used and disseminated; and the development of the ecosystem needed to promote growth of AI and ML in cancer imaging.

Radiology historically has been a leader of digital transformation in healthcare. The introduction of digital imaging systems, picture archiving and communication systems (PACS), and teleradiology transformed radiology services over the past 30 years. Radiology is again at the crossroad for the next generation of transformation, possibly evolving as a one-stop integrated diagnostic service. Artificial intelligence and machine learning promise to offer radiology new powerful new digital tools to facilitate the next transformation. The radiology community has been developing computer-aided diagnosis (CAD) tools based on machine learning (ML) over the past 20 years. Among various AI techniques, deep-learning convolutional neural networks (CNN) and its variants have been widely used in medical image pattern recognition. Since the 1990s, many CAD tools and products have been developed. However, clinical adoption has been slow due to a lack of substantial clinical advantages, difficulties integrating into existing workflow, and uncertain business models. This paper proposes three pathways for AI's role in radiology beyond current CNN based capabilities 1) improve the performance of CAD, 2) improve the productivity of radiology service by AI-assisted workflow, and 3) develop radiomics that integrate the data from radiology, pathology, and genomics to facilitate the emergence of a new integrated diagnostic service.

We developed a DICOM dataset that can be used to evaluate the performance of de-identification algorithms. DICOM objects (a total of 1,693 CT, MRI, PET, and digital X-ray images) were selected from datasets published in the Cancer Imaging Archive (TCIA). Synthetic Protected Health Information (PHI) was generated and inserted into selected DICOM Attributes to mimic typical clinical imaging exams. The DICOM Standard and TCIA curation audit logs guided the insertion of synthetic PHI into standard and non-standard DICOM data elements. A TCIA curation team tested the utility of the evaluation dataset. With this publication, the evaluation dataset (containing synthetic PHI) and de-identified evaluation dataset (the result of TCIA curation) are released on TCIA in advance of a competition, sponsored by the National Cancer Institute (NCI), for algorithmic de-identification of medical image datasets. The competition will use a much larger evaluation dataset constructed in the same manner. This paper describes the creation of the evaluation datasets and guidelines for their use.

Longitudinal personal health record (PHR) provides a foundation for managing patients' health care, but we do not have such a system in the U.S. except for the patients in the Department of Veterans Affairs. Such a gap exists mainly in the rest of the U.S. by the fact that patients' electronic health records are scattered across multiple health care facilities and often not shared due to privacy, security, and business interests concerns from both patients and health care organizations. In addition, patients have ethical concerns related to consent. To patients, data security, privacy, and consent are based on trustfulness, rather than patients’ engagement in ensuring only authorized people can view their PHRs with patient-managed granularity. Resolving these challenges is an important step in making longitudinal PHR useful for patient care. This research aims to design and implement a blockchain-enabled sharing platform prototype for PHR with desired patient-controlled data security, privacy, and consent granularity. Built upon our prior work of a blockchain-enabled access control (BAC) model, we design a blockchain-enabled sharing platform for PHR with patient-controlled security, privacy, and consent granularity. We further implement the construct by building a prototypical platform among a patient and two typical health care organizations. Health organizations that hold the patient's electronic health records can join the platform with trust based on the validation from the patient. The mutual trust can be established through a rigorous validation process by both the patient and the built-in Hyperledger Fabric blockchain consensus mechanism. We proposed a system trusted by patients and health care providers and constructed a Web-based PHR sharing platform with patient-controlled security, privacy, and consent granularity. We analyzed the system scalability in three aspects and showed millisecond range of performance when simultaneously changing access permissions on hundreds of PHRs. Consent, security and privacy of the model are ensured by the merits of the BAC model. We discovered the current blockchain model limits the system scalability due to using a non-graphical database. A new graphical database is suggested for future improvements. In this research, we report a solution to electronically sharing and managing patients’ electronic health records originating from multiple organizations, focusing on privacy, security, and granularity control of consent in the U.S. Specifically, the system protects data security and privacy, and provides auditability, scalability, distributedness, patient consent autonomy, and zero-trust capabilities. The prototypical instantiation of the designed model suggested the feasibility of combining emerging blockchain technology with next generation access control model to tackle a longstanding longitudinal PHR problem.

In recent years, many computer-aided diagnosis schemes have been proposed to assist radiologists in detecting lung nodules. The research efforts have been aimed at increasing the sensitivity while decreasing the false-positive detections on digital chest radiographs. Among the problems of reducing the number of false positives, the differentiation between nodules and end-on vessels is one of the most challenging tasks performed by computer. Most investigators have used a conventional two-stage pattern recognition approach, ie, feature extraction followed by feature classification. The performance of this approach depends totally on good feature definition in the feature extraction stage. Unfortunately, suitable feature definition and corresponding extraction implementation algorithms proved to be very difficult to define and specify. A convolution neural network (CNN) architecture, trained by direct connection to the raw image is proposed to tackle the problem. The CNN, which uses locally responsive activation function, is directly and locally connected to the raw image. The performance of the CNN is evaluated in comparison to an expert radiologist. We used the receiver operating characteristics (ROC) method with area under the curve (A_z) as the performance index to evaluate all the simulation results. The CNN showed superior performance (A_z=0.99) to the radiologist’s (A_z=0.83). The CNN approach can potentially be applied to other applications, such as the differentiation of film defects and microcalcifications in mammography, in which the image features are difficult to define or not known a priori.

The authors address radiology image management and communication systems (IMAC) in the context of hospital-wide patient care activities. Some features of radiology imaging operations are discussed along with a specific example of the image capable network now under development at Georgetown University Medical Center. Some detail and specific data on operational aspects of radiology service within a hospital are also described.<>