Normal Workflow Research Articles

AI upscaling Supporting Image Alignment in Photogrammetric Reconstruction

Abstract. This research aims to investigate and analyze the contribution provided by preventive image processing using AI in the SfM data acquisition process. Specifically, the objective is to observe qualities and defects of "AI upscaling" integrated into the normal workflow of digital restitution, with the hypothesis that greater sharpness and resolution can lead to better alignments and better model generations. Other similar experiments have been carried out previously on the AI intervention in the photos to improve the alignments, but a generic procedure and with untrained public AI has not yet been experimented. In the first phase of the research, the most suitable AI upscaling method for the purpose is selected, processing example images and comparing them with the original and various settings of the AI parameters to maintain likeness. In the second phase, the procedure efficiency was evaluated in several ways: through the automatic evaluation of Agisoft Metashape's Image Quality, by directly comparing the resulting model with the standard digital reconstruction of the same objects, comparing with a model produced with the native AI upscaling of the return program, and finally comparing with a high-level survey without defects. This test was repeated for four case studies specifically diversified by several factors, including subject scale, level of detail, lighting, color, blur, and type of camera used. Clear improvements emerge in all cases of AI upscaling, even reaching successful reconstructions that would not have started with the canonical method. It can therefore be stated that this preventive upscaling procedure is decisive in cases of low quality or damaged datasets, while in cases of already qualitatively sufficient photos it can be a useful support for increasing the level of detail.

Impact of an Oncology Clinical Pharmacist Intervention on Clinical Trial Enrollment in The US Oncology Network’s MYLUNG Consortium

Introduction: The Molecularly Informed Lung Cancer Treatment in a Community Cancer Network: A Pragmatic Consortium™ (MYLUNG) clinical trial platform aims to advance the use of precision medicine in patients with non–small cell lung cancer through a series of prospective and iterative clinical trials. Timely patient accrual onto oncology clinical trials is a known practice challenge and impaired accrual rates can lead to premature trial closure or properly powered trial outcomes. The US Oncology Network recently implemented a clinical pharmacist (ClinReview) initiative to provide remote clinical services to screen patients for enrollment onto MYLUNG Protocol 2. This study aims to evaluate the effect of the remote clinical pharmacist intervention on study enrollment rates. Methods: An oncology-trained clinical pharmacist remotely reviewed systemic chemotherapy treatment orders during normal workflow and, in addition, a weekly custom recruitment report within six community Network practices (149 physicians). The pharmacist identified, screened, and assisted with the communication regarding eligible patients for enrollment. The onsite research team received timely and relevant patient data to facilitate expedited enrollment. Enrollment and intervention data were tracked to monitor the impact of the pharmacist intervention. Monthly enrollment was evaluated using a paired t-test. Results: Over 8 months, the pharmacist screened 506 potentially eligible patients; 34% were enrolled. Average monthly enrollment was significantly greater following the ClinReview intervention (3.4 vs. 6.6 patients/month; p = .02). Among the 289 patients not enrolled, 73% exceeded their eligibility window, 9% died or enrolled into hospice, 4% declined participation, and 13% transferred care or were treated at outside facilities. Conclusions: Incorporating an oncology clinical pharmacist into the clinical research team was associated with improved clinical trial enrollment. Validation of the effect of multidisciplinary interventions across a broader spectrum of differentially resourced oncology practices will be conducted within future MYLUNG iterations.

Radiologically derived 3D virtual models for neurosurgical planning

Three dimensional (3D) virtual models for neurosurgery have demonstrated substantial clinical utility, especially for neuro-oncological cases. Computer-aided design (CAD) modelling of radiological images can provide realistic and high-quality 3D models which neurosurgeons may use pre-operatively for surgical planning. 3D virtual models are useful as they are the basis for other models that build off this design. 3D virtual models are quick to segment but can also be easily added to normal neurosurgical and radiological workflow without disruption. Three anatomically complex neuro-oncology cases that were referred from a single institution by three different neurosurgeons were segmented and 3D virtual models were created for pre-operative surgical planning. A face-to-face interview was performed with the surgeons after the models were delivered to gauge the usefulness of the model in pre-surgical planning. All three neurosurgeons found that the 3D virtual model was useful for presurgical planning. Specifically, the virtual model helped in planning operative positioning, understanding spatial relationship between lesion and surrounding critical anatomy and identifying anatomy that will be encountered intra-operatively in a sequential manner. It provided benefit in Multidisciplinary team (MDT) meetings and patient education for shared decision making.3D virtual models are beneficial for pre-surgical planning and patient education for shared decision making for neurosurgical neuro-oncology cases. We believe this could be further expanded to other surgical specialties. The integration of 3D virtual models into normal workflow as the initial step will provide an easier transition into modalities that build off the virtual models such as printed, virtual, augmented and mixed reality models.

Development of a Cancer Rehabilitation Dashboard to Collect Data on Physical Function in Cancer Patients and Survivors.

The aim of the study is to describe the development of a cancer rehabilitation dashboard that collects data on physical function for cancer survivors in a cancer institute. This project was conducted at the Miami Cancer Institute. The cancer rehabilitation dashboard was developed by a team of physicians, biostatistician, and medical informatics teams to record, report and track the physical function of cancer survivors. A multimodal approach to the measurement of physical function was used and included the Patient-Reported Outcome Measurement Information System-Physical Function short form, Patient-Reported Outcome Measurement Information System-Fatigue short form, Timed Up and Go Test, Sit-to-Stand Test in 30-sec test, four-stage balance test, and grip strength. To develop this system, a Cerner Power Form was developed based on the physical function data. To display the data, a dedicated flowsheet was developed and placed within the Oncology Viewpoint in Cerner Millennium. Thus, from inside any patient record, the flowsheet could easily be accessed by providers without leaving normal clinician workflows. Using native functionality, the data can also be shown in graphical format to facilitate dialog with patients and oncology teams. All patient data from the Cerner Power Form discrete task assays were integrated into an existing Oncology Data Warehouse for all patients. The data elements in the Cerner Power Form were identified in the electronic medical record system, loaded into the Oncology Data Warehouse, and related to the other source systems to develop reports and data visualizations such as the cancer rehabilitation dashboard. The cancer rehabilitation dashboard allows visualization of numerous parameters of physical function in cancer survivors evaluated and treated and their change over time. Rendered in Tableau, the cancer rehabilitation dashboard acts as a centralized, interactive data source to analyze and connect clinicians to near real-time data. The cancer rehabilitation dashboard was successfully developed and implemented into a cancer rehabilitation practice in a cancer institute and used to collect and track physical function data for cancer survivors receiving treatment and cancer survivors. This information has been used to direct the treatment plan and educate individual patients about the impact of the cancer and its treatment on physical function as well as oncology teams in a cancer institute. The cancer rehabilitation dashboard provides an insight into the physical function of cancer survivors receiving treatment and cancer survivors using both self-reported and objective metrics. It can be customized to suit the interests of clinicians and researchers wishing to improve the quality of life of this population.

Safety in Health Care: The Impact of Operating Room Design.

The science of operating room design has grown over the past 20years due to the realization that the physical environment influences health care provider performance and patient outcomes. Medical errors occur when the normal workflow in an operating room is disrupted as providers must overcome sub-optimal conditions. All aspects of the physical environment can impact operating room flow. Studying the layout, contents, ergonomics, and environmental parameters of the operating can lead improved work conditions resulting improved patient and provider safety. At the forefront of operating room design science is the use of simulation and the evaluation of new technologies.

Exception Reporting: A Novel Approach to Tracking of Multiple Behavioral Health Outcomes.

Federally funded medical and behavioral healthcare programs often have substantial evaluation outcome tracking and reporting requirements, which can become burdensome to program staff resulting in decreased buy-in, increased chance of staff burnout and turnover, and less rigorous and consistent data collection efforts. To address this issue, a novel data collection approach, "exception reporting," was implemented to supplement and support the required data collection for a federally funded Assertive Outpatient Treatment (AOT) program. This work details the process and outcomes related to exception reporting for this comprehensive behavioral health treatment program that serves justice involved clients with serious mental illness (SMI). Results indicate that exception reporting was easily integrated into clinician's normal workflows and resulted in a number of benefits. Specifically, results indicated that exception reporting decreased the data collection burden for program staff while allowing them to efficiently track program outcomes required by the funder. Additional research into which practice settings exception reporting can most easily be integrated into, and which client outcomes may be best tracked using this methodology, is indicated.

A Simple and Fast Method for the Formation and Downstream Processing of Cancer-Cell-Derived 3D Spheroids: An Example Using Nicotine-Treated A549 Lung Cancer 3D Spheres.

Although 2D in vitro cancer cell cultures have been used for decades as a first line-of-research tool to investigate antitumoral drugs and treatments, their use presents many drawbacks, including the poor resemblance of such cultures to the characteristics of in vivo tumors. To mitigate these drawbacks, 3D culture models have emerged as a more representative alternative. Cancer cells cultured as 3D structures have the advantage of resembling solid tumors in their architecture and in their resistance to chemotherapeutic drugs, in part because of restrained drug penetration. Additionally, these 3D structures create a more physiological environment for the study of immune cell invasion and migration, comparable to solid tumors. In this paper, we describe a fast and cost-effective step-by-step protocol for the generation of 3D spheres using ultra-low-attachment (ULA) multiwell plates, which can be incorporated into the normal workflow of any laboratory. Using this protocol, spheroids of different human cancer cell lines can be obtained and can then be characterized on the basis of their morphology, viability, and expression of specific markers.

B-137 Development of a Novel Software System for Providing Real-Time Laboratory Test Information to Clinicians at Point of Order Entry

Abstract Background Clinicians depend on an increasingly complex set of laboratory tests to aid in the diagnosis and monitoring of patients. Staying informed about the proper usage of these tests requires active effort, especially as test offerings continually evolve. To aid in test decision making, institutions often implement some form of decision support, including: grouping of common “order sets”, “best practice alert” popups, and web-based “lab catalogs”. For many institutions, lab catalogs provide the most comprehensive information on available tests, but typically exist separate from the interface used to place test orders, and thus suffer from underutilization in actual clinical practice. Methods We developed a novel, open-source software to closely integrate our institution’s lab catalog at the point of order entry within our electronic medical record (EMR) system. The software provides an immediate pop-up window anytime a clinician needs additional information while placing orders. The software also provides an independent, customizable search function that helps steer clinicians away from commonly mis-ordered tests. Both functions are immediately available at the point of order entry, without requiring the user to leave the EMR or otherwise interrupt their normal workflow. We conducted a small usability pilot with a focus group of clinicians, using the System Usability Scale (SUS) (1). Results Our software reduced the number of clicks and keystrokes needed to access detailed lab information from 27 to 2. The focus group (n = 6) rated the product favorably on the SUS, with average rating 89 ± 9 out of 100. Conclusion This study demonstrates proof-of-concept for a novel software decision support tool that provides instant access to lab information at the point of order entry. Future work will focus on evaluating utility in a real-world clinical pilot.

Normal Workflow and Key Strategies for Data Cleaning Toward Real-World Data: Viewpoint.

With the rapid development of science, technology, and engineering, large amounts of data have been generated in many fields in the past 20 years. In the process of medical research, data are constantly generated, and large amounts of real-world data form a "data disaster." Effective data analysis and mining are based on data availability and high data quality. The premise of high data quality is the need to clean the data. Data cleaning is the process of detecting and correcting "dirty data," which is the basis of data analysis and management. Moreover, data cleaning is a common technology for improving data quality. However, the current literature on real-world research provides little guidance on how to efficiently and ethically set up and perform data cleaning. To address this issue, we proposed a data cleaning framework for real-world research, focusing on the 3 most common types of dirty data (duplicate, missing, and outlier data), and a normal workflow for data cleaning to serve as a reference for the application of such technologies in future studies. We also provided relevant suggestions for common problems in data cleaning.

Radiology E-Consults: Reinterpretation of Outside CT and MRI Exams to Improve Quality, Reduce Unneeded Imaging, and Lower Costs

ObjectiveAdvanced imaging examinations commonly uncover incidental findings. Without access to prior imaging that is external to their healthcare system, radiologists may struggle to assess the significance of these findings. This can lead to unnecessary imaging follow-up recommendations and management uncertainty. We created a program to facilitate outside advanced imaging reinterpretation in the context of both internal and external prior imaging, resulting in more complete and accurate follow-up recommendations. MethodsOur value-based care organization created a program to facilitate outside advanced imaging reinterpretation, including importing the outside examinations and reports into our PACS/RIS platform. A total of 222 of these reinterpretations were performed between November 2020 and June 2022. In each case, a radiologist reinterpreted the entire examination and placed an official report into the electronic medical record, comparing their conclusions and recommendations with those of the outside imaging report. The reinterpretation report included a trackable code for future data mining purposes. ResultsThese E-consults seamlessly fitted into a radiologist's normal workflow and resulted in revenue generation as well as wRVU credit. The E-consults resulted in a change in management recommendations for approximately 46% of the overall number of cases. Approximately 37% of all examinations resulted in a recommendation of no further imaging required, or adjustment of the imaging modality/interval. DiscussionAn official radiology E-consult program can synthesize disparate sources of imaging data, transforming it into clear and documented recommendations which providers can use to improve patient safety and reduce unnecessary imaging and costs.

Fully immersive virtual reality for skull-base surgery: surgical training and beyond.

A virtual reality (VR) system, where surgeons can practice procedures on virtual anatomies, is a scalable and cost-effective alternative to cadaveric training. The fully digitized virtual surgeries can also be used to assess the surgeon's skills using measurements that are otherwise hard to collect in reality. Thus, we present the Fully Immersive Virtual Reality System (FIVRS) for skull-base surgery, which combines surgical simulation software with a high-fidelity hardware setup. FIVRS allows surgeons to follow normal clinical workflows inside the VR environment. FIVRS uses advanced rendering designs and drilling algorithms for realistic bone ablation. A head-mounted display with ergonomics similar to that of surgical microscopes is used to improve immersiveness. Extensive multi-modal data are recorded for post-analysis, including eye gaze, motion, force, and video of the surgery. A user-friendly interface is also designed to ease the learning curve of using FIVRS. We present results from a user study involving surgeons with various levels of expertise. The preliminary data recorded by FIVRS differentiate between participants with different levels of expertise, promising future research on automatic skill assessment. Furthermore, informal feedback from the study participants about the system's intuitiveness and immersiveness was positive. We present FIVRS, a fully immersive VR system for skull-base surgery. FIVRS features a realistic software simulation coupled with modern hardware for improved realism. The system is completely open source and provides feature-rich data in an industry-standard format.

Implementing a workflow-integrated motivational interviewing training program for psychiatry trainees on an inpatient consultation-liaison rotation: lessons learned.

Effective consultation-liaison psychiatry (CLP) is proactive, collaborative, and requires providers to have proficiency with therapeutic skills beyond nosology and medication management. Motivational interviewing (MI) is an evidenced-based intervention that should be considered essential for CLP trainees to learn. Given that the demands of training and patient care are already experienced as stressful for many psychiatry trainees, the authors endeavored to create a MI training program that was integrated into trainees' normal CLP workflow. Twenty-two trainees on an inpatient CLP rotation participated in a six-week MI training program that was incorporated into their regular workflow. The program included didactic sessions with role-playing, as well as on-demand between-session coaching via an expert in MI. Trainee participation and perceptions of MI were measured via a questionnaire that was administered prior to each training session. Trainee participation in the didactic sessions was inconsistent. Questionnaire data revealed positive baseline perceptions of motivational interviewing and its usefulness in inpatient medical settings. Additionally, as trainees participated in the program, perceived knowledge of motivational interviewing as well as awareness of motivational issues among their patients increased. Finally, participation in program was not perceived as disruptive to daily workflow for the participants. This the first documented attempt at implementing a MI training program for CLP trainees that was integrated into their regular workflow. Preliminary data identified some encouraging trends, but also unexpected challenges. These lessons could inform how these types of training programs are implemented moving forward.

A High-Speed Hyperspectral Laparoscopic Imaging System.

Minimally invasive surgery (MIS) has expanded broadly in the field of abdominal and pelvic surgery. Laparoscopic and robotic surgery has improved surgeon ergonomics, instrument precision, operative time, and postoperative recovery across various abdominal procedures. The goal of this study is to establish the feasibility of implementing high-speed hyperspectral imaging into a standard laparoscopic setup and exploring its benefit to common intracorporeal procedures. A hyperspectral laparoscopic imaging system was constructed using a customized hyperspectral camera alongside a standard rigid laparoscope and was validated for both spectral and spatial accuracy. Demosaicing methods were investigated for improved full-resolution visualization. Hyperspectral cameras with different spectral ranges were considered and compared with one another alongside two different light sources to determine the most effective configuration. Finally, different porcine tissues were imaged ex-vivo to test the capabilities of the system and spectral footprints of the various tissues were extracted. The tissue was also imaged in a phantom to simulate the system's use in MIS. The results demonstrated a hyperspectral laparoscopic imaging system that could provide quantitative, diagnostic information while not disrupting normal workflow nor adding excessive weight to the laparoscopic setup. The high-speed hyperspectral laparoscopic imaging system can have immediate applications in image-guided surgery.

Clinical Implementation of a Combined Artificial Intelligence and Natural Language Processing Quality Assurance Program for Pulmonary Nodule Detection in the Emergency Department Setting.

This quality assurance study assessed the implementation of a combined artificial intelligence (AI) and natural language processing (NLP) program for pulmonary nodule detection in the emergency department setting. The program was designed to function outside of normal reading workflows to minimize radiologist interruption. In all, 19,246 CT examinations including at least some portion of the lung anatomy performed in the emergent setting from October 1, 2021, to June 1, 2022, were processed by the combined AI-NLP program. The program used an AI algorithm trained on 6-mm to 30-mm pulmonary nodules to analyze CT images and an NLP to analyze radiological reports. Cases flagged as negative for pulmonary nodules by the NLP but positive by the AI algorithm were classified as suspected discrepancies. Discrepancies result in secondary review of examinations for possible addenda. Out of 19,246 CT examinations, 50 examinations (0.26%) resulted in secondary review, and 34 of 50 (68%) reviews resulted in addenda. Of the 34 addenda, 20 patients received instruction for new follow-up imaging. Median time to addendum was 11 hours. The majority of reviews and addenda resulted from missed pulmonary nodules on CT examinations of the abdomen and pelvis. A background quality assurance process using AI and NLP helped improve the detection of pulmonary nodules and resulted in increased numbers of patients receiving appropriate follow-up imaging recommendations. This was achieved without disrupting in-shift radiologist workflow or causing significant delays in patient follow for the diagnosed pulmonary nodule.

Integrated Genomic DNA/RNA Profiling vs Fluorescence in Situ Hybridization in the Detection of MYC and BCL2 (and BCL6) Rearrangements in Large B-Cell Lymphomas: Updates Amid the New WHO Classification of Lymphoid Neoplasms.

Large B-cell lymphomas (LBCLs) are a heterogeneous group of lymphoid neoplasms whose molecular and cytogenetic profile has predictive and prognostic implications. The concept of double-hit lymphomas (DHLs) was recently updated in the fifth edition of the World Health Organization classification, with the exclusion of MYC and BCL6 rearranged tumors from the group. Now, DHLs are referred to as diffuse large B-cell lymphoma/high-grade B-cell lymphoma with MYC and BCL2 rearrangements. Fluorescence in situ hybridization (FISH) is the current gold standard for detecting rearrangements in LBCLs, but comprehensive genomic profiling (CGP) has recently been suggested to be at least as accurate as FISH in classifying these neoplasms and providing additional genetic information. We analyzed a cohort of 131 patients in whom FISH and CGP studies were performed as part of our normal clinical workflow and compared the effectiveness of FISH and CGP in detecting these clinically relevant rearrangements. Our findings are in agreement with our previously published study, which analyzed a cohort of 69 patients, supporting our hypothesis that the best approach to maximize detection of DHLs while limiting waste seems to be a combination of CGP and MYC break-apart FISH testing, the latter to capture the presence of non-IGH::MYC events. Our study supports the combined use of FISH and GCP rather than either method alone to better detect MYC and BCL2 (and BCL6) gene rearrangements.

Effect of forwarder multipassing on forest soil parameters changes

In the lowland part of Croatia, heavy machinery such as forwarders is mainly used for the purpose of extracting wood from even-aged forest stands. According to the forest management plan, forwarders are used intensively in the winter period when the soil is mostly saturated with water and when their activity can cause significant damage to the soil. The aim of this study was to determine changes in soil characteristics as a consequence of the repeated passage of a loaded 8-wheel forwarder on silty clay loam type of soil. The research was conducted in an area where the forwarder usually works and in a way that did not significantly disrupt his normal workflow. The results indicate that during the study period the soil had a good bearing capacity and that the observed changes in soil characteristics (bulk density, total soil porosity, soil moisture, particle density, soil water retention capacity etc.) occurred as a result of breaking structural soil aggregates after soil compaction by multiple passes. Characteristic points (T) of equalized penetration curves indicate the compaction of the soil surface layer. Cone penetration index (CI) values did not show a proportional increase as the number of forwarder passes increased, although significant differences in their values with respect to the number of passes were found. Shear strength (τ) did not significantly increase with increasing the number of passes, but a statistically significant difference in the measured values was detected at the soil surface, which was not observed at a depth of 15 cm. Exceeding the rut depth limit of 10 cm occurred only after the 20th pass. Our results indicate that the soil at the harvesting site had a good bearing capacity during the study period.

A hybrid method to improve efficiency of patient specific SRS and SBRT QA using 3D secondary dose verification.

Patient Specific QA (PSQA) by direct phantom measurement for all intensity modulated radiation therapy (IMRT) cases is labor intensive and an inefficient use of the Medical Physicist's time. The purpose of this work was to develop a hybrid quality assurance (QA) technique utilizing 3D dose verification as a screening tool to determine if a measurement is necessary. This study utilized Sun Nuclear DoseCHECK (DC), a 3D secondary verification software, and Fraction 0, a trajectory log IMRT QA software. Twenty-two Lung stereotactic body radiation therapy (SBRT) and thirty single isocentre multi-lesion SRS (MLSRS) plans were retrospectively analysed in DC. Agreement of DC and the TPS dose for selected dosimetric criteria was recorded. Calculated 95% confidence limits (CL) were used to establish action limits. All cases were delivered and measured using the Sun Nuclear stereotactic radiosurgery (SRS) MapCheck. Trajectory logs of the delivery were used to calculate Fraction 0 results for the same criteria calculated by DC. Correlation of DC and Fraction 0 results were calculated. Phantom measured QA was compared to Fraction 0 QA results for the cases which had DC criteria action limits exceeded. Correlation of DC and Fraction 0 results were excellent, demonstrating the same action limits could be used for both and DC can predict Fraction 0 results. Based on the calculated action limits, zero lung SBRT cases and six MLSRS cases were identified as requiring a measurement. All plans that passed the DC screening had a passing measurement based PSQA and agreed with Fraction 0 results. Using 95% CL action limits of dosimetric criteria, a 3D secondary dose verification can be used to determine if a measurement is required for PSQA. This method is efficient for it is part of the normal clinical workflow when verifying any clinical treatment. In addition, it can drastically reduce the number of measurements needed for PSQA.

Eye-Tracking Active Indicators of Insider Threats: Detecting Illicit Activity During Normal Workflow

Cyber challenges faced by organizations today involve malicious inside actors, often labeled insider threats (ITs). These present a difficult challenge in that the most well-designed cybersecurity apparatus is vulnerable to those inside the organization who have privileged access to information systems. Innovative methods must be developed to help security analysts narrow the large pool of potential ITs in large organizations to a more manageable number. The purpose of this article is to develop and validate eye-tracking metrics that are diagnostic of IT behavior. Key stimuli, or called active indicator probes, were embedded into a simulated workflow to elicit diagnostic eye-tracking responses. Two environments were simulated: financial and intelligence analysis. We evaluated participants performing as regular workers relative to ITs to identify metrics that distinguished between the two groups. Detection of illicit eye gaze behavior while using chat programs was possible when conversations with accomplices occurred in a separate chat window from normal permissible chat conversations. Validation of results in real work environments is necessary for practical application. However, if the approach proves to translate successfully, automated monitoring of eye-tracking responses may augment existing insider detection methods, within frameworks for best practices in organizational security and cyber defense.

RUNX1 Mutated AML and MDS: Similarities, Differences and Molecular Factors Leading to Disease Progression

Background: Mutations in the myeloid transcription factor RUNX1 are frequently detected in AML and MDS. Differences have been described regarding additional cytogenetic and molecular genetic abnormalities, however the molecular factors responsible for progression of RUNX1 mutated MDS to AML are yet not fully elucidated. Aim: Analysis of genetic (1) similarities and (2) differences of RUNX1 mutated AML and MDS cases, (3) identification of factors potentially causing disease progression from RUNX1 mutated MDS to AML. Methods: Whole-genome sequencing (WGS) was performed for 773 AML and 747 MDS patients (median coverage 100x). 151bp paired-end reads were generated on NovaSeq 6000 and HiSeqX machines (Illumina, San Diego, CA). As no sample specific normal tissue was available, a so-called Tumor/Unmatched normal (TUN) workflow was used to reduce technical artefacts and germline calls. All reported p-values are two-sided and were considered significant at p<0.05. Results: Mutations in RUNX1 (RUNX1mut) were detected at comparable frequencies in AML and MDS (AML: 97/773 cases, 13%; MDS: 71/747, 10%). In MDS, an association of the RUNX1mut frequency with higher blast counts was observed (0-5% blasts: 15/419 with RUNX1mut, 4%; ≥5,<10%: 16/175, 9%; ≥10,<20%: 40/153, 26%). However, blast count did not influence overall survival (OS) in RUNX1mut subgroups (0-5% blasts: 18 months (m); ≥5,<10%: 25 m; ≥10,<20%: 18 m; AML: 21 m). The variant allele frequency (VAF) of RUNX1mut differed significantly dependent on the blast count (median VAF in cases <5% blasts: 0.22; ≥5%: 0.42; p<0.001). Regarding cytogenetic abnormalities, RUNX1mut MDS showed a higher frequency of normal karyotype (NK) compared to AML (66% vs. 38%, p<0.001) but similar frequencies of trisomies (18% vs. 22%) and complex karyotype (3% vs. 5%), although differences regarding the trisomy type were detected (MDS: only trisomy 8; AML: trisomies 8, 11 and 13). In 10% of AML cases with RUNX1mut a rearrangement involving MECOM was detected (5% GATA::MECOM, 5% other MECOM rearrangements), which was absent in MDS with RUNX1mut. Similarly, KMT2A-PTD and FLT3-ITD alterations were detected in AML but not in MDS with RUNX1mut (KMT2A-PTD: 7% in AML vs. 0% in MDS, p=0.021; FLT3-ITD: 13% vs. 0%, p=0.001). By contrast, TP53mut were found at comparably low frequencies (MDS: 4/71, 6%; AML: 3/97, 3%). For both AML and MDS, a strong association of RUNX1mut with mutations in spliceosome genes (SF3B1, SRSF2, U2AF1, ZRSR2) was detected (AML: 58/97 = 60% in RUNX1mut vs. 139/676 = 21% in RUNX1wild type (wt); p<0.001; MDS: 52/71 = 73% vs. 346/676 = 51%, p<0.001). However, while SRSF2mut in MDS were associated with increasing blast counts similar to RUNX1mut (0-5% blasts: 25/419 with SRSF2mut, 6%; ≥5,<10%: 130/175, 17%; ≥10,<20%: 50/153, 32%), SF3B1mut were negatively correlated to blast counts and thus to RUNX1mut (0-5% blasts: 179/419, 43%; ≥5,<10%: 28/175, 16%; ≥10,<20%: 11/153, 7%). In the total AML cohort, in 24/773 of cases a CN-LOH involving RUNX1 was detected, 21 of these also showed a RUNX1mut (22% (21/97) of RUNX1mut cases). By contrast, in the total MDS cohort in only 2/747 cases a CN-LOH covering RUNX1 was found, none of them harboring a RUNX1mut. The VAF of cases with RUNX1mut + CN-LOH (RUNX1mutLOH) was >0.5 for all cases and was significantly higher than the VAF of RUNX1mut cases without CN-LOH (0.89 (range 0.54-0.97) vs. 0.42 (range 0.05-0.85), p<0.001). Moreover, cases with RUNX1mutLOH were found to be significantly associated with FLT3-ITD (RUNX1mutLOH, 6/21 = 29%; RUNX1mut without CN-LOH, 7/76 = 9%; p=0.032). Conclusions: (1) RUNX1mut occur at similar frequencies in AML and MDS and genetically show an MDS signature, reflected e.g. by association with mutations in spliceosome genes. OS is similar in all subgroups independent of blast counts. (2) In MDS, RUNX1mut are associated with higher blast counts, thus potentially functioning as driver towards progression to AML. (3) Additional cytogenetic/molecular genetic alterations that might function as progression factors from RUNX1mut MDS to AML include: (i) karyotype, especially MECOM rearrangements, trisomy 11 and trisomy 13 (ii) KMT2A-PTD, (iv) FLT3-ITD, (v) and CN-LOH covering RUNX1 (Figure). (4) By contrast, TP53mut do not seem to play a role in progression of RUNX1 mutated MDS to AML. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Automated Gating and Interpretation of Clinical Flow Cytometry Data: A Computational Approach using Artificial Intelligence and Deep Learning

Abstract Flow cytometry (FCM) allows pathologists to accurately immunophenotype hematopoietic cells by detecting the expression of surface proteins with fluorochrome-conjugated antibodies. Accordingly, FCM plays a pivotal role in the diagnostic workup of many hematologic malignancies. However, post-analytic processing and the interpretation of FCM data are primarily manual processes that impact the consistency and limit the throughput of the method. The post-analytic processing workflow is colloquially referred to as ‘gating’ and involves the identification and characterization of immune cell populations by hand through successions of biaxial plots. The gated FCM data is then used by the pathologist to make a clinical interpretation and to then determine potential diagnoses from the patterns and cell frequencies seen in bivariate plots of the data. Since gating and interpretation are manually performed, post-analytic analyses of FCM data are not only laborious from a workflow perspective but remain subjective and prone to variability based on the experience and skill of both the medical laboratory scientist and pathologist. The objective of this study was to develop a computational pipeline that leverages machine learning-based (ML) solutions to automate gating and clinical interpretation of FCM data to increase the throughput and improve the repeatability of FCM analysis. Raw FCS files from clinical samples being evaluated for the presence of T-cell lymphoproliferative disease were exported from the on-instrument database. Automated gating was performed using open-source, supervised-ML packages for flow cytometry data (flowCore and flowDensity; R). These packages implement all processing steps that would typically be done manually (e.g. applying compensation, quality control (QC), and gating). Plots of interest were then generated from the gated data and classified as normal or abnormal using the clinical interpretations that were applied during the normal clinical workflow. These binary labels were used to train an ML-based classifier (VGG-19; Python). To evaluate pipeline performance, we collected FCS files from 1,188 samples that were analyzed by our flow cytometry lab. The automated gating pipeline was used to gate for CD4+CD3+ cells and to create bivariate plots for CD7/CD26 expression. 1150 (96.8%) passed QC and were all gated correctly by visual inspection. Of the 38 (3.2%) samples that failed QC, 13 (1.1%) were lymphopenic and were gated correctly by visual inspection, and the remaining 25 (2.1%) were gated manually. Using the CD7/CD26 plots, the classifier demonstrated promising predictive performance and achieved a precision of 0.85 and recall of 0.83 (weighted average across classes). Our findings represent a novel effort to automate both the gating and interpretation of FCM data using artificial intelligence. These results suggest that ML-based tools have potential utility in aiding the processing and interpretation of FCM data and can augment the efficiency and consistency of workflows in the clinical flow cytometry laboratory.