Interpretable & Explainable Machine Learning for Ultrasonic Defect Sizing

Despite its popularity in literature, there are few examples of machine learning (ML) being used for industrial nondestructive evaluation (NDE) applications. A significant barrier is the 'black box' nature of most ML algorithms. This paper aims to improve the interpretability and explainability of ML for ultrasonic NDE by presenting a novel dimensionality reduction method: Gaussian feature approximation (GFA). GFA involves fitting a 2D elliptical Gaussian function an ultrasonic image and storing the seven parameters that describe each Gaussian. These seven parameters can then be used as inputs to data analysis methods such as the defect sizing neural network presented in this paper. GFA is applied to ultrasonic defect sizing for inline pipe inspection as an example application. This approach is compared to sizing with the same neural network, and two other dimensionality reduction methods (the parameters of 6 dB drop boxes and principal component analysis), as well as a convolutional neural network applied to raw ultrasonic images. Of the dimensionality reduction methods tested, GFA features produce the closest sizing accuracy to sizing from the raw images, with only a 23% increase in RMSE, despite a 96.5% reduction in the dimensionality of the input data. Implementing ML with GFA is implicitly more interpretable than doing so with principal component analysis or raw images as inputs, and gives significantly more sizing accuracy than 6 dB drop boxes. Shapley additive explanations (SHAP) are used to calculate how each feature contributes to the prediction of an individual defect's length. Analysis of SHAP values demonstrates that the GFA-based neural network proposed displays many of the same relationships between defect indications and their predicted size as occur in traditional NDE sizing methods.

Exploring the potential of using simulation games for engaging with sheep farmers about lameness recognition

Introduction

Computer simulation games are increasingly being used in agriculture as a promising tool to study, support and influence real-life farming practices. We explored the potential of using simulation games to engage with sheep farmers on the ongoing challenge of reducing lameness. Working with UK stakeholders, we developed a game in which players are challenged with identifying all the lame sheep in a simulated flock. Here, we evaluate the game's potential to act as a tool to help assess, train and understand farmers' ability to recognize the early signs of lameness.

Methods

Participants in the UK were invited to play the game in an online study, sharing with us their in-game scores alongside information relating to their real-life farming experience, how they played the game, and feedback on the game. Mixed methods were used to analyze this information in order to evaluate the game. Quantitative analyses consisted of linear modeling to test for statistical relationships between participants' in-game recall (% of the total number of lame sheep that were marked as lame), and the additional information they provided. Qualitative analyses of participants' feedback on the game consisted of thematic analysis and a Likert Scale questionnaire to contextualize the quantitative results and identify additional insights from the study.

Results

Quantitative analyses identified no relationships between participants' (n = 63) recall scores and their real life farming experience, or the lameness signs they looked for when playing the game. The only relationship identified was a relationship between participants' recall score and time spent playing the game. Qualitative analyses identified that participants did not find the game sufficiently realistic or engaging, though several enjoyed playing it and saw potential for future development. Qualitative analyses also identified several interesting and less-expected insights about real-life lameness recognition practices that participants shared after playing the game.

Discussion

Simulation games have potential as a tool in livestock husbandry education and research, but achieving the desired levels of realism and/or engagingness may be an obstacle to realizing this. Future research should explore this potential further, aided by larger budgets and closer collaboration with farmers, stockpeople, and veterinarians.

Uncertainty Quantification for Deep Learning in Ultrasonic Crack Characterization

Deep learning for nondestructive evaluation (NDE) has received a lot of attention in recent years for its potential ability to provide human level data analysis. However, little research into quantifying the uncertainty of its predictions has been done. Uncertainty quantification (UQ) is essential for qualifying NDE inspections and building trust in their predictions. Therefore, this article aims to demonstrate how UQ can best be achieved for deep learning in the context of crack sizing for inline pipe inspection. A convolutional neural network architecture is used to size surface breaking defects from plane wave imaging (PWI) images with two modern UQ methods: deep ensembles and Monte Carlo dropout. The network is trained using PWI images of surface breaking defects simulated with a hybrid finite element / ray-based model. Successful UQ is judged by calibration and anomaly detection, which refer to whether in-domain model error is proportional to uncertainty and if out of training domain data is assigned high uncertainty. Calibration is tested using simulated and experimental images of surface breaking cracks, while anomaly detection is tested using experimental side-drilled holes and simulated embedded cracks. Monte Carlo dropout demonstrates poor uncertainty quantification with little separation between in and out-of-distribution data and a weak linear fit ( R=0.84 ) between experimental root-mean-square-error and uncertainty. Deep ensembles improve upon Monte Carlo dropout in both calibration ( R=0.95 ) and anomaly detection. Adding spectral normalization and residual connections to deep ensembles slightly improves calibration ( R=0.98 ) and significantly improves the reliability of assigning high uncertainty to out-of-distribution samples.

Domain Adapted Deep-Learning for Improved Ultrasonic Crack Characterization Using Limited Experimental Data

Deep learning is an effective method for ultrasonic crack characterization due to its high level of automation and accuracy. Simulating the training set has been shown to be an effective method of circumventing the lack of experimental data common to nondestructive evaluation (NDE) applications. However, a simulation can neither be completely accurate nor capture all variability present in the real inspection. This means that the experimental and simulated data will be from different (but related) distributions, leading to inaccuracy when a deep learning algorithm trained on simulated data is applied to experimental measurements. This article aims to tackle this problem through the use of domain adaptation (DA). A convolutional neural network (CNN) is used to predict the depth of surface-breaking defects, with in-line pipe inspection as the targeted application. Three DA methods across varying sizes of experimental training data are compared to two non-DA methods as a baseline. The performance of the methods tested is evaluated by sizing 15 experimental notches of length (1-5 mm) and inclined at angles of up to 20° from the vertical. Experimental training sets are formed with between 1 and 15 notches. Of the DA methods investigated, an adversarial approach is found to be the most effective way to use the limited experimental training data. With this method, and only three notches, the resulting network gives a root-mean-square error (RMSE) in sizing of 0.5 ± 0.037 mm, whereas with only experimental data the RMSE is 1.5 ± 0.13 mm and with only simulated data it is 0.64 ± 0.044 mm.

Multi-modality imaging in hypertrophic cardiomyopathy: intermodal discrepancies in key prognostic parameters

Background/Introduction

Multi-modality imaging is crucial for confirming diagnosis and assessing prognosis in patients with hypertrophic cardiomyopathy (HCM). However, inter-modality discrepancies in key parameters are commonly reported.

Purpose

To assess real-world inter-modal reporting discrepancies between transthoracic echocardiography (TTE) and cardiac magnetic resonance (CMR) imaging in the measurement of four key parameters in HCM patients.

Methods

Consecutive HCM patients with TTE and CMR performed within 6 months of each other at a tertiary centre were retrospectively assessed for reported maximum wall thickness (MWT), left atrial diameter (LAd), left ventricular ejection fraction (LVEF) and presence of left ventricular apical aneurysm (LVAA). The CMR report was considered gold standard. Data are reported as mean ± standard deviation (SD) or median and interquartile range (IQR) as appropriate.

Results

353 consecutive HCM patients (72% male, median age 60.9 years, IQR 49.8–71.6 years) with TTE and CMR within 6 months (median difference 1.7 months, IQR 1.1–3.4 months) were assessed between 4th January 2018 and 9th April 2019. Of 284 patients with paired MWT data, median difference was 0.0 mm (IQR −1.0 to 3.0 mm, p=0.02), likely representing a difference in distributions of MWT. TTE both over and underestimated MWT (in 36% and 46% cases respectively).

Of the 94 patients with paired LAd data, mean difference was 0.4±5.7 mm (95% CI −0.8010 to 1.546, p=0.5).

N=320 patients with paired LVEF data (after excluding patients with atrial fibrillation (n=20)). Median difference in LVEF was 12% (IQR 5–19% p<0.0001). TTE underestimated LVEF in 88% of cases. CMR and TTE both identified 14 (5%) patients as having LVEF <50%. There were however 8 cases of disagreement in classification of LVEF <50%, due to over (n=4) or underestimation (n=4) by TTE.

LVAA was accurately identified by TTE in only 9/30 (30%) of those patients with demonstrable LVAA by CMR (p=0.0008). TTE evidence of a discreet apical chamber (paradoxical jet on spectral or colour Doppler) was present in 16/21 (76%) cases where TTE failed to overtly identify LVAA. However, apical or mid-cavity obliteration was reported in 15/21 (71%) cases where TTE failed to identify LVAA.

Conclusion(s)

Echocardiography and CMR measurements are often used interchangeably in clinical practice but inter-modality discrepancies can affect diagnosis and sudden cardiac death (SCD) risk assessment. This is particularly important for binary risk factors such as LVEF<50% or LVAA which are considered major SCD risk factors in the latest American Heart Association guidelines. 25 (7%) patients in our cohort had major risk factors identified by CMR that were not identified on TTE. CMR is an important, recommended tool where TTE imaging is suboptimal, but attention to more subtle elements of abnormal intracavity blood flow may be able to increase LVAA detection during TTE.

Funding Acknowledgement

Type of funding sources: None.

Deep Learning for Ultrasonic Crack Characterization in NDE

Machine learning for nondestructive evaluation (NDE) has the potential to bring significant improvements in defect characterization accuracy due to its effectiveness in pattern recognition problems. However, the application of modern machine learning methods to NDE has been obstructed by the scarcity of real defect data to train on. This article demonstrates how an efficient, hybrid finite element (FE) and ray-based simulation can be used to train a convolutional neural network (CNN) to characterize real defects. To demonstrate this methodology, an inline pipe inspection application is considered. This uses four plane wave images from two arrays and is applied to the characterization of cracks of length 1-5 mm and inclined at angles of up to 20° from the vertical. A standard image-based sizing technique, the 6-dB drop method, is used as a comparison point. For the 6-dB drop method, the average absolute error in length and angle prediction is ±1.1 mm and ±8.6°, respectively, while the CNN is almost four times more accurate at ±0.29 mm and ±2.9°. To demonstrate the adaptability of the deep learning approach, an error in sound speed estimation is included in the training and test set. With a maximum error of 10% in shear and longitudinal sound speed, the 6-dB drop method has an average error of ±1.5 mmm and ±12°, while the CNN has ±0.45 mm and ±3.0°. This demonstrates far superior crack characterization accuracy by using deep learning rather than traditional image-based sizing.

Peritoneal cytology as a prognostic indicator in endometrial carcinoma

Among the prognostic indicators of endometrial carcinoma, the presence of cancer cells in peritoneal washings has been suggested as indicating a poor prognosis. This proposition was examined for 373 patients; positive peritoneal washings were observed in 20 of 335 patients with stage I disease and 10 of 29 with stages II-IV. In 13 of the 30 patients there was no gross evidence of disease. Within a follow-up period of two years, positive cytology appeared to be a sign of poor prognosis.

Adenoid cystic carcinoma of the cervix: a report of 17 cases

Seventeen cases of adenoid cystic carcinoma of the cervix are presented. The long-term survival was in early stage IB-IIA. Patients with more than stage IIB had no survival rate over 5 years. One thing of interest was the fact that lung metastasis was 29.4%.

Clear cell carcinoma of the endometrium

The clinical data of 22 patients with clear cell adenocarcinoma of the endometrium treated at the University of North Carolina Memorial Hospital are reported. In addition, the data with particular reference to survival, site of recurrence, and treatment are combined with information from two previous reports of clear cell adenocarcinoma of the endometrium to better define survival. It is noted that the patients with clear cell adenocarcinoma of the endometrium were older and had an overall poorer survival than is reported for adenocarcinoma of the endometrium (nonclear cell). Patients with Stage I clear cell carcinoma of the endometrium, however, had a similar five-year survival to Stage I adenocarcinoma of the endometrium. The paper also examines treatment methods and correlates these with site of recurrence as well as survival.

Lactate production in McArdle's disease

A case of McArdle's disease in a man is described in detail and a less complete study of his family is reported. This patient showed the classical features of McArdle's disease and the diagnosis was confirmed by muscle biopsy. Unlike other reported cases of this disorder, this case showed a normal rise in blood lactate levels on ischaemic exercise. This apparently paradoxical finding is discussed. It is suggested that a normal rise in the level of blood lactate on ischaemic exercise should not exclude myophosphorylase deficiency.

Myophosphorylase deficiency (McArdle's disease) in two interrelated families

The clinical and laboratory findings are presented of three patients (two affected sisters and their male cousin) with myophosphorylase deficiency in two interrelated families. Nine unaffected relatives were also investigated. Although the three patients demonstrated the characteristic features of the disease, their unaffected relatives showed no clear evidence of a heterozygous state. The genetic findings support the hypothesis that the disease is inherited as a rare autosomal recessive. A possible sex-limited mode of inheritance is discussed.

Oligodendroglioma in an infant of eight months

Images