J. Imaging, Volume 6, Issue 11 (November 2020) – 16 articles

The classification of the musculoskeletal images can be very challenging, mostly when it is being done in the emergency room, where a decision must be made rapidly. The computer vision domain has gained increasing attention in recent years, due to its achievements in image classification. The convolutional neural network (CNN) is one of the latest computer vision algorithms that achieved state-of-the-art results. A CNN requires an enormous number of images to be adequately trained, and these are always scarce in the medical field. Transfer learning is a technique that is being used to train the CNN by using fewer images. In this paper, we study the appropriate method to classify musculoskeletal images by transfer learning and by training from scratch. We applied six state-of-the-art architectures and compared their performance with transfer learning and with a network trained from scratch. From our results, transfer learning did increase the model performance significantly, and, additionally, it made the model less prone to overfitting. Full article

Baggage travelling on a conveyor belt in the sterile area (the rear collector located after the check-in counters) often gets stuck due to traffic jams, mainly caused by incorrect entries from the check-in counters on the collector belt. Using suitcase appearance captured on the Baggage Handling System (BHS) and airport checkpoints and their re-identification allows for us to handle baggage safer and faster. In this paper, we propose a Siamese Neural Network-based model that is able to estimate the baggage similarity: given a set of training images of the same suitcase (taken in different conditions), the network predicts whether the two input images belong to the same baggage identity. The proposed network learns discriminative features in order to measure the similarity among two different images of the same baggage identity. It can be easily applied on different pre-trained backbones. We demonstrate our model in a publicly available suitcase dataset that outperforms the leading latest state-of-the-art architecture in terms of accuracy. Full article

Background: The aim of this work is to identify an automatic, accurate, and fast deep learning segmentation approach, applied to the parenchyma, using a very small dataset of high-resolution computed tomography images of patients with idiopathic pulmonary fibrosis. In this way, we aim to enhance the methodology performed by healthcare operators in radiomics studies where operator-independent segmentation methods must be used to correctly identify the target and, consequently, the texture-based prediction model. Methods: Two deep learning models were investigated: (i) U-Net, already used in many biomedical image segmentation tasks, and (ii) E-Net, used for image segmentation tasks in self-driving cars, where hardware availability is limited and accurate segmentation is critical for user safety. Our small image dataset is composed of 42 studies of patients with idiopathic pulmonary fibrosis, of which only 32 were used for the training phase. We compared the performance of the two models in terms of the similarity of their segmentation outcome with the gold standard and in terms of their resources’ requirements. Results: E-Net can be used to obtain accurate (dice similarity coefficient = 95.90%), fast (20.32 s), and clinically acceptable segmentation of the lung region. Conclusions: We demonstrated that deep learning models can be efficiently applied to rapidly segment and quantify the parenchyma of patients with pulmonary fibrosis, without any radiologist supervision, in order to produce user-independent results. Full article

In digital neutron imaging, the neutron scintillator screen is a limiting factor of spatial resolution and neutron capture efficiency and must be improved to enhance the capabilities of digital neutron imaging systems. Commonly used neutron scintillators are based on ⁶LiF and gadolinium oxysulfide neutron converters. This work explores boron-based neutron scintillators because ¹⁰B has a neutron absorption cross-section four times greater than ⁶Li, less energetic daughter products than Gd and ⁶Li, and lower γ-ray sensitivity than Gd. These factors all suggest that, although borated neutron scintillators may not produce as much light as ⁶Li-based screens, they may offer improved neutron statistics and spatial resolution. This work conducts a parametric study to determine the effects of various boron neutron converters, scintillator and converter particle sizes, converter-to-scintillator mix ratio, substrate materials, and sensor construction on image quality. The best performing boron-based scintillator screens demonstrated an improvement in neutron detection efficiency when compared with a common ⁶LiF/ZnS scintillator, with a 125% increase in thermal neutron detection efficiency and 67% increase in epithermal neutron detection efficiency. The spatial resolution of high-resolution borated scintillators was measured, and the neutron tomography of a test object was successfully performed using some of the boron-based screens that exhibited the highest spatial resolution. For some applications, boron-based scintillators can be utilized to increase the performance of a digital neutron imaging system by reducing acquisition times and improving neutron statistics. Full article

The contactless estimation of vital signs using conventional color cameras and ambient light can be affected by motion artifacts and changes in ambient light. On both these problems, a multimodal 3D imaging system with an irritation-free controlled illumination was developed in this work. In this system, real-time 3D imaging was combined with multispectral and thermal imaging. Based on 3D image data, an efficient method was developed for the compensation of head motions, and novel approaches based on the use of 3D regions of interest were proposed for the estimation of various vital signs from multispectral and thermal video data. The developed imaging system and algorithms were demonstrated with test subjects, delivering a proof-of-concept. Full article

(1) The objective of the present study was to identify suitable parameters to determine the (degree of) freshness of Bell pepper fruit of three colors (yellow, red, and green) over a two-week period including the occurrence of shrivel using non-destructive real-time measurements (2) Materials and methods: Surface glossiness was measured non-destructively with a luster sensor type CZ-H72 (Keyence Co., Osaka, Japan), a colorimeter, a spectrometer and a profilometer type VR-5200 (Keyence) to obtain RGB images. (3) Results: During storage and shelf life, bell pepper fruit of initially 230–245 g lost 2.9–4.8 g FW per day at 17 °C and 55% rh. Shriveling started at 6–8% weight loss after 4–5 days and became more pronounced. Glossiness decreased from 450–500 a.u. with fresh fruit without shrivel, 280–310 a.u. with moderately shriveled fruit to 80–90 a.u. with severely shriveled fruit irrespective of color against a background of <40 a.u. within the same color, e.g., light red and dark red. Non-invasive color measurements showed no decline in Lab values (chlorophyll content), irrespective of fruit color and degree of shrivel. RGB images, converted into false color images, showed a concomitant increase in surface roughness (Sa) from Sa = ca. 2 µm for fresh and glossy, Sa = ca. 7 µm for moderately shriveled to Sa = ca. 24 µm for severely shriveled rough surfaces of stored pepper fruit, equivalent to a 12-fold increase in surface roughness. The light reflectance peak at 630–633 nm was universal, irrespective of fruit color and freshness. Hence, a freshness index based on (a) luster values ≥ 450 a.u., (b) Sa ≤ 2 µm and (c) the difference in relative reflectance in % between 630 nm and 500 nm is suggested. The latter values declined from ca. 40% for fresh red Bell pepper, ca. 32% after 6 days when shriveling had started, to ca. 21% after 12 days, but varied with fruit color. (4) Conclusion: overall, it can be concluded that color measurements were unsuitable to determine the freshness of Bell pepper fruit, whereas profilometer, luster sensor, and light reflectance spectra were suitable candidates as a novel opto-electronic approach for defining and parametrizing fruit freshness. Full article

Deep learning algorithms have become the first choice as an approach to medical image analysis, face recognition, and emotion recognition. In this survey, several deep-learning-based approaches applied to breast cancer, cervical cancer, brain tumor, colon and lung cancers are studied and reviewed. Deep learning has been applied in almost all of the imaging modalities used for cervical and breast cancers and MRIs for the brain tumor. The result of the review process indicated that deep learning methods have achieved state-of-the-art in tumor detection, segmentation, feature extraction and classification. As presented in this paper, the deep learning approaches were used in three different modes that include training from scratch, transfer learning through freezing some layers of the deep learning network and modifying the architecture to reduce the number of parameters existing in the network. Moreover, the application of deep learning to imaging devices for the detection of various cancer cases has been studied by researchers affiliated to academic and medical institutes in economically developed countries; while, the study has not had much attention in Africa despite the dramatic soar of cancer risks in the continent. Full article

A majority of foodborne illnesses result from inappropriate food handling practices. One proven practice to reduce pathogens is to perform effective hand-hygiene before all stages of food handling. In this paper, we design a multi-camera system that uses video analytics to recognize hand-hygiene actions, with the goal of improving hand-hygiene effectiveness. Our proposed two-stage system processes untrimmed video from both egocentric and third-person cameras. In the first stage, a low-cost coarse classifier efficiently localizes the hand-hygiene period; in the second stage, more complex refinement classifiers recognize seven specific actions within the hand-hygiene period. We demonstrate that our two-stage system has significantly lower computational requirements without a loss of recognition accuracy. Specifically, the computationally complex refinement classifiers process less than 68% of the untrimmed videos, and we anticipate further computational gains in videos that contain a larger fraction of non-hygiene actions. Our results demonstrate that a carefully designed video action recognition system can play an important role in improving hand hygiene for food safety. Full article

Artworks have a layered structure subjected to alterations caused by various factors. The monitoring of defects at sub-millimeter scale may be performed by laser interferometric techniques. The aim of this work was to develop a compact system to perform laser speckle imaging in situ for effective mapping of subsurface defects in paintings. The device was designed to be versatile with the possibility of optimizing the performance by easy parameters adjustment. The system exploits a laser speckle pattern generated through an optical diffuser and projected onto the artworks and image correlation techniques for the analysis of the speckle intensity pattern. A protocol for the optimal measurement was suggested, based on calibration curves for tuning the mean speckle size in the acquired intensity pattern. The system was validated in the analysis of detachments in an ancient painting model using a short pulse thermal stimulus to induce a surface deformation field and standard decorrelation algorithms for speckle pattern matching. The device is equipped with a compact thermal camera for preventing any overheating effects during the phase of the stimulus. The developed system represents a valuable nondestructive tool for artwork diagnostics, allowing the monitoring of subsurface defects in paintings in out-of-laboratory environment. Full article

The moisture content of screed samples is an essential parameter in the construction industry, since the screed must dry to a certain level of moisture content to be ready for covering. This paper introduces neutron radiography (NR) and neutron tomography (NT) as new, non-destructive techniques for analysing the drying characteristics of screed. Our NR analyses evaluate the results of the established methods while offering much higher spatial resolution of 200 $\mathsf{\mu }$m, thereby facilitating a two- and three-dimensional understanding of screed’s drying behaviour. Because of NR’s exceptionally high sensitivity regarding the total cross section of hydrogen the precise moisture content of screed samples is obtainable, resulting in new observations. The current methods to measure moisture content comprise the ‘calcium carbide method’, the ‘Darr method’, and electrical sensor systems. Full article

This paper proposes a performance model for estimating the user time needed to transcribe small collections of handwritten documents using a keyword spotting system (KWS) that provides a number of possible transcriptions for each word image. The model assumes that only information obtained from a small training set is available, and establishes the constraints on the performance measures to achieve a reduction of the time for transcribing the content with respect to the time required by human experts. The model is complemented with a procedure for computing the parameters of the model and eventually estimating the improvement of the time to achieve a complete and error-free transcription of the documents. Full article

Daltonisation refers to the recolouring of images such that details normally lost by colour vision deficient observers become visible. This comes at the cost of introducing artificial colours. In a previous work, we presented a gradient-domain colour image daltonisation method that outperformed previously known methods both in behavioural and psychometric experiments. In the present paper, we improve the method by (i) finding a good first estimate of the daltonised image, thus reducing the computational time significantly, and (ii) introducing local linear anisotropic diffusion, thus effectively removing the halo artefacts. The method uses a colour vision deficiency simulation algorithm as an ingredient, and can thus be applied for any colour vision deficiency, and can even be individualised if the exact individual colour vision is known. Full article

Face-morphing operations allow for the generation of digital faces that simultaneously carry the characteristics of two different subjects. It has been demonstrated that morphed faces strongly challenge face-verification systems, as they typically match two different identities. This poses serious security issues in machine-assisted border control applications and calls for techniques to automatically detect whether morphing operations have been previously applied on passport photos. While many proposed approaches analyze the suspect passport photo only, our work operates in a differential scenario, i.e., when the passport photo is analyzed in conjunction with the probe image of the subject acquired at border control to verify that they correspond to the same identity. To this purpose, in this study, we analyze the locations of biologically meaningful facial landmarks identified in the two images, with the goal of capturing inconsistencies in the facial geometry introduced by the morphing process. We report the results of extensive experiments performed on images of various sources and under different experimental settings showing that landmark locations detected through automated algorithms contain discriminative information for identifying pairs with morphed passport photos. Sensitivity of supervised classifiers to different compositions on the training and testing sets are also explored, together with the performance of different derived feature transformations. Full article

In spectral-spatial classification of hyperspectral image tasks, the performance of conventional morphological profiles (MPs) that use a sequence of structural elements (SEs) with predefined sizes and shapes could be limited by mismatching all the sizes and shapes of real-world objects in an image. To overcome such limitation, this paper proposes the use of object-guided morphological profiles (OMPs) by adopting multiresolution segmentation (MRS)-based objects as SEs for morphological closing and opening by geodesic reconstruction. Additionally, the ExtraTrees, bagging, adaptive boosting (AdaBoost), and MultiBoost ensemble versions of the extremely randomized decision trees (ERDTs) are introduced and comparatively investigated for spectral-spatial classification of hyperspectral images. Two hyperspectral benchmark images are used to validate the proposed approaches in terms of classification accuracy. The experimental results confirm the effectiveness of the proposed spatial feature extractors and ensemble classifiers. Full article

In order to tackle three-dimensional tumor volume reconstruction from Positron Emission Tomography (PET) images, most of the existing algorithms rely on the segmentation of independent PET slices. To exploit cross-slice information, typically overlooked in these 2D implementations, I present an algorithm capable of achieving the volume reconstruction directly in 3D, by leveraging an active surface algorithm. The evolution of such surface performs the segmentation of the whole stack of slices simultaneously and can handle changes in topology. Furthermore, no artificial stop condition is required, as the active surface will naturally converge to a stable topology. In addition, I include a machine learning component to enhance the accuracy of the segmentation process. The latter consists of a forcing term based on classification results from a discriminant analysis algorithm, which is included directly in the mathematical formulation of the energy function driving surface evolution. It is worth noting that the training of such a component requires minimal data compared to more involved deep learning methods. Only eight patients (i.e., two lung, four head and neck, and two brain cancers) were used for training and testing the machine learning component, while fifty patients (i.e., 10 lung, 25 head and neck, and 15 brain cancers) were used to test the full 3D reconstruction algorithm. Performance evaluation is based on the same dataset of patients discussed in my previous work, where the segmentation was performed using the 2D active contour. The results confirm that the active surface algorithm is superior to the active contour algorithm, outperforming the earlier approach on all the investigated anatomical districts with a dice similarity coefficient of 90.47 ± 2.36% for lung cancer, 88.30 ± 2.89% for head and neck cancer, and 90.29 ± 2.52% for brain cancer. Based on the reported results, it can be claimed that the migration into a 3D system yielded a practical benefit justifying the effort to rewrite an existing 2D system for PET imaging segmentation. Full article

This paper presents a unique approach for the dichotomy between useful and adverse variations of key-point descriptors, namely the identity and the expression variations in the descriptor (feature) space. The descriptors variations are learned from training examples. Based on labels of the training data, the equivalence relations among the descriptors are established. Both types of descriptor variations are represented by a graph embedded in the descriptor manifold. Invariant recognition is then conducted as a graph search problem. A heuristic graph search algorithm suitable for the recognition under this setup was devised. The proposed approach was tested on the FRGC v2.0, the Bosphorus and the 3D TEC datasets. It has shown to enhance the recognition performance, under expression variations, by considerable margins. Full article