Using Traditional Machine Learning Techniques for Predicting the Best Clinical Outcomes On Improvement In Vas, Mod And Ncos Based Upon Clinical And Imaging Features

Using Traditional Machine Learning Techniques for Predicting the Best Clinical Outcomes On Improvement In Vas, Mod And Ncos Based Upon Clinical And Imaging Features

PURPOSE OR LEARNING OBJECTIVE:

To use machine learning techniques for predicting the clinical outcomes on improvement in VAS, MOD, and NCOS based on the managements like Spinal decompression without fusion (open discectomy/Laminectomy), Spinal compression with fusion, and Conservative Management.

METHODS OR BACKGROUND:

Patients with symptomatic lumbar spine disease with back pain with or without radiculopathy and neurological deficit were enrolled. The primary outcome measures were Visual analogue scale (VAS), Modified Oswestry Disability Index (MOD), and Neurogenic Claudication Outcome Score (NCOS) collected at pre-operatively and at 3 months post-operatively. The further analysis studied the following factors to determine if any are predictive of outcomes: sex, BMI, occupation, involvement in sports, herniation type, depression, work status, herniation level, duration of symptoms, and history of past spine surgery. The features were selected and machine learning models were trained to predict the improvement in the primary outcome measures. The results were evaluated on the basis of the ROC-AUC score for different classes.

RESULTS OR FINDINGS:

There were a total of 200 entries of patients with Lumbar Spine Disease between age 18 and above. Among the various Machine Learning Models, Random Forest Classifier gave the best ROC-AUC score in all three classes. The AUC score for VAS, MOD, and NCOS was 0.877, 0.8215, and 0.830 respectively and the macro-average AUC score was found to be 0.84 (see Fig.1).

 Fig 1: ROC-AUC scores of different Machine Learning Algorithms on Test Dataset.

 

CONCLUSION:

Machine Learning model could be used as a predictive tool for deciding the type of management that a patient should undergo to achieve the best results. Based on the predicted improvement in different indices for the particular management type, the predictions could help Surgeons for deciding the type of management that would be most beneficial for the patient.

Estimation of bias of deep learning-based chest X-ray classification algorithm

Estimation Of Bias Of Deep Learning-Based Chest X-Ray Classification Algorithm​

PURPOSE OR LEARNING OBJECTIVE:

To evaluate the bias in the diagnostic performance of a deep learning-based chest X-ray classification algorithm on previously unseen external data.

METHODS OR BACKGROUND:

632 chest X-rays were randomly collected from an academic centre hospital and anonymised selectively, leaving out fields needed for the bias estimation (manufacturer name, age, and gender). They were from six different vendors – AGFA (388), Carestream (45), DIPS (21), GE (31), Philips (127), and Siemens (20). The male and female distribution was 376 and 256. The X-rays were read for consolidation ground truth establishment on CARING analytics platform (CARPL). These X-rays were run on open-sourced chest X-ray classification model. Inferencing results were analysed using Aequitas, an open-source python-based package to detect the presence of bias, fairness of algorithms. Algorithms’ performance was evaluated on the three metadata classes – gender, age group, and brand of equipment. False omission rate (FOR) and false-negative rate (FNR) metrics were used for calculating the inter-class scores of bias.

RESULTS OR FINDINGS:

AGFA, 60 to 80 age group, and male were the dominant entities and hence considered as baseline for evaluation of bias towards other classes. Significant false omission rate (FOR) and false negative rate (FNR) disparities were observed for all vendor classes except Siemens as compared to AGFA. No gender disparity was seen. All groups show FNR parity whereas all classes showed disparity with respect to false omission rate for age.

CONCLUSION:

We demonstrate that AI algorithms may develop biases, based on the composition of training data. We recommend bias evaluation check to be an integral part of every AI project. Despite this, AI algorithms may still develop certain biases, some of those difficult to evaluate.

LIMITATIONS:

Limited pathological classes were evaluated.

Why Standardisation Of Pre-inferencing Image Processing Methods Is Crucial For Deep Learning Algorithms – A Compelling Evidence Based On The Variations In Outputs For Different Inferencing Workflow

Why Standardisation Of Pre-Inferencing Image Processing Methods Is Crucial For Deep Learning Algorithms – A Compelling Evidence Based On The Variations In Outputs For Different Inferencing Workflow​

PURPOSE OR LEARNING OBKECTIVE:

To evaluate if there are statistically significant differences in the outputs of a deep learning algorithm on two inferencing workflows, with different unit-processing methods.

METHODS OR BACKGROUND:

The study was performed on DeepCOVIDXR, an open-source algorithm for the detection of COVID19 on chest X-rays. It is an ensemble of convolutional neural networks developed to detect COVID-19 on frontal chest radiographs. The algorithm was evaluated using a dataset of 905 Chest X-rays containing 484 COVID+ cases (as determined RTPCR test) and 421 COVID negative cases. The algorithm supports both batch image processing (workflow1) and single image processing (workflow2) for running inferencing. All the Xray were inferenced using both methods. In batch image processing, images were resized (224×224 and 331×331) and then lung was cropped out, but in single image processing, cropping was done without resizing of images.

RESULTS OR FINDINGS:

We observed a significant difference in the results for the two inferencing workflows. The AUC for COVID classification was 0.632 on the bulk image processing pathway whereas it was 0.769 for the single image processing. There were discordant results in 334 studies, 164 were classified as positive in workflow1 whereas negative in workflow2 whereas 170 X-rays that were classified as negative on workflow1 were classified as positive in workflow2.

CONCLUSION:

We report statistically significant differences in the results of a deep learning algorithm on using different inferencing workflows.

LIMITATIONS:

With rising adoption of radiology AI, it is important to understand that seemingly innocuous changes in the processing pathways can led to disastrous results in the clinical results.

Enhanced separation of brain tumors and edema via diffusion tensor distribution imaging: Illustration with lymphoma cases

Enhanced Separation Of Brain Tumors And Edema Via Diffusion Tensor Distribution Imaging: Illustration With Lymphoma Cases

PURPOSE OR LEARNING OBJECTIVE:

To investigate the clinical potential of diffusion tensor distribution imaging (DTD) for visually differentiating brain tumors and edema from healthy tissue non-invasively.

METHODS:

Multidimensional diffusion (MDD) MRI images were acquired in 2 lymphoma patients on a 3T Discovery 750w system (GE Healthcare) with a 32-channel head coil. Prototype spin-echo EPI sequences were performed using the following parameters: TR/TE=3298/121 ms, in-plane resolution = 3×3 mm2. MDD consisted of 43 linear and 37 spherical b-tensors at b = 100, 700, 1400, 2000 s/mm2. Total scan time was ~5 min. Post-processing of the data was done using dVIEWR powered by MICE Toolkit (www.dviewr.com). The main features related to average cell density (mean diffusivity, MD) and cell elongation (microscopic anisotropy) can be computed within “bins” corresponding to specific tissue types, i.e., “thin” for elongated cells (e.g., white matter), “thick” for densely packed round cells (e.g., grey matter), “sparse” for low cell-density diffusion environments (e.g., edema) and “big” for free water (e.g., ventricles).

RESULTS:

Bin-resolved segmentation maps (SegM) facilitate the identification of edematous regions, captured by the sparse bin (red areas in SegM). These regions surround the investigated lymphomas, themselves mostly captured by the thin bin (green in SegM), indicating that they consist of elongated cells. These cells are randomly oriented, as they appear white (red+green+blue) in the thin-bin mean-orientation maps (see Figure 1). The bin-resolved MD maps’ colors highlight the inverse relationship between MD and average cell density across different tissue types. In particular, the sparse bin exhibits an intermediate MD characteristic of edema.


Figure 1. Diffusion Tensor Distribution (DTD) parameter maps of lymphoma cases.


CONCLUSION:

DTD could provide enhanced visualization tools for radiologists aiming to better separate/characterize healthy and pathological tissues non-invasively.

LIMITATIONS:

This pilot study had limitation in terms of small sample size.

Implementation Of Fast Echo-planar Imaging (EPImix) MRI Sequence For Scan Time Reduction In Critical And Unco-operative Patients

Implementation Of Fast Echo-Planar Imaging (EPImix) MRI Sequence For Scan Time Reduction In Critical And Unco-Operative Patients​

PURPOSE OR LEARNING OBJECTIVE:

To detail how a fast multi-contrast Echo planar Image mix (EPIMix) MRI sequence can lead to a successful reduction in scan time in critical and uncooperative patients compared to the routine clinical brain imaging without compromising the adequate image quality and diagnosis.

METHODS:

A prospective pilot study was conducted on 29 patients requiring emergent brain imaging for concerns of stroke(3), tremors(2), slurring of speech(3), headache(6), memory loss(4), imbalance(2), limb weakness(6), aphasia(1), dementia(1) and Parkinson\’s disease(1) using EPIMix brain imaging sequence on the Discovery 750w 3T, GE Healthcare MR system. EPIMix brain MRI consisting of six contrasts (T2*, T1/T2-FLAIR, T2, DWI, ADC) was acquired in 72-75 seconds. Routine T1w/T2w axial, coronal FLAIR and T2w sagittal images were also concurrently acquired and were correlated with EPIMix images for all the patients. Qualitative analysis of the EPIMix scans was performed by two experienced radiologists for assessment of diagnostic accuracy, artifacts, and image quality.

RESULTS:

The image quality was diagnostic in all of these cases (100%) and the diagnostic performance was comparable between EPIMix and routine clinical MRI without much significant difference, indicating the preservation of adequate image quality on fast EPIMix scans (see Fig.1).



Fig 1. (i) 74-year-old male presented with a history of slurred speech. There is a chronic infarct with gliosis in the right parietal region. The internal content shows hyperintensity on T2WI (A) and T2-FLAIR (B), and hypointensity on T1-FLAIR (D)(arrows); no diffusion restriction on DWI (F) is seen (arrows). (ii) 71-year old male presented with a history of upper limb tremors. Hyperintensity in the right frontal periventricular white matter is seen on T2WI (G) and T2-FLAIR (H) and hypointensity on T1-FLAIR (J)(arrows) with reduced size of the frontal horn, possibly due to ependymitis granularis; no diffusion restriction on DWI is seen to suggest acute ischaemia (L) (arrows). (iii) 82-year-old male presented with a clinical profile of stroke. Cortical & subcortical gliosis is seen in the left middle frontal gyrus. The internal content shows hyperintensity on T2WI (M) and T2-FLAIR (N), and hypointensity on T1-FLAIR (P)(arrows); no diffusion restriction on DWI (R) is seen(arrows).

CONCLUSION:

The pilot study reveals that the EPIMix sequence with rapid scanning can minimize motion artifacts and can be used in unstable patients to evaluate a wide range of brain pathologies without compromising diagnostic image quality.

LIMITATIONS:

EPIMix produces six weighted MRI contrasts in a short time, albeit some image artifacts such as geometric distortion at the skull base and susceptibility artifacts, which were noticed in almost all EPIMix scans. Image degradation with the above-mentioned artifacts is the result of an inherent trade-off between scan time reduction and image quality.