Machine learning monitoring for laser osteotomy

This work proposes a new online monitoring method for an assistance during laser osteotomy. The method allows differentiating the type of ablated tissue and the applied dose of laser energy. The setup analyzes the laser-induced acoustic emission, detected by an airborne microphone sensor. The analysis of the acoustic signals is carried out using a machine learning algorithm that is pre-trained in a supervised manner. The efficiency of the method is experimentally evaluated with several types of tissues, which are: skin, fat, muscle, and bone. Several cutting-edge machine learning frameworks are tested for the comparison with the resulting classification accuracy in the range of 84–99%. It is shown that the datasets for the training of the machine learning algorithms are easy to collect in real-life conditions. In the future, this method could assist the doctors during laser osteotomy, minimizing the damage of the nearby healthy tissues and provide cleaner pathologic tissue removal.     

Front cover

Continuous-wave terahertz reflection imaging is a potential tool for biological tissues. Based on our home-made continuouswave terahertz reflection imaging system, the effect of both polarization mode and reflection window on the imaging performance is studied theoretically and experimentally, showing good agreement. By taking obtaining sample information and image contrast into consideration, p-polarized terahertz waves are recommended. Moreover, considering the sample boundary identification and the image contrast, selection criteria for reflection window are proposed. This work will help to improve the performance of continuous-wave terahertz reflection imaging and accelerate the THz imaging in biological application. Further details can be found in the article by Limin Wu, Yuye Wang, Haibin Li, Zelong Wang, Meilan Ge, Degang Xu, and Jianquan Yao ( e202100245 ).

     

Polarization clustering of biological structures with Mueller matrix parameters

Mueller matrix imaging polarimetry (MMIP) is a promising technique for the characterization of biological tissues, including the classification of microstructures in pathological diagnosis. To expand the parameter space of Mueller matrix parameters, we propose new vector parameters (VPs) according to the Mueller matrix polar decomposition method. We measure invasive bladder cancer (IBC) with extensive necrosis and high-grade ductal carcinoma in situ (DCIS) with MMIP, and the regions of cancer cells and fibrotic stroma are classified with the VPs. Then the proposed and existing VPs are mapped on the Poincaré sphere with 3D visualization, and an indicator of spatial feature is defined based on the minimum enclosing sphere to evaluate the classification capability of the VPs. For both IBC and DCIS, the results show that the proposed VPs exhibit evident contrast between the regions of cancer cells and fibrotic stroma. This study broadens the fundamental Mueller matrix parameters and helps to improve the characterization ability of the MMIP technique.     

Inside Cover

A high‐efficiency computer‐aided diagnostic model of ovarian cancer was developed, integrating SHG imaging technology for non‐invasive imaging of living tissue and machine learning method based on radiomics and TPOT. This model can rapidly, non‐destructively, and accurately perform ovarian cancer diagnosis and has great potential in improving diagnostic efficacy and efficiency of medical pathologists. Further details can be found in the article by Guangxing Wang, Yang Sun, Youting Chen, Qiqi Gao, Dongqing Peng, Hongxin Lin, Zhenlin Zhan, Zhiyi Liu, and Shuangmu Zhuo ( e202000050 ).

     

Prediction of radiotherapy response in cervix cancer by Raman spectroscopy: a pilot study

Radiotherapy is the choice of treatment for locally advanced stages of the cervical cancers, one of the leading female cancers. Because of intrinsic factors, tumors of same clinical stage and histological type often exhibit differential radioresponse. Radiotherapy regimen, from first fraction of treatment to clinical evaluation of response, spans more than 4 months. Clinical assessment by degree of tumor shrinkage is the only routinely practiced method to evaluate the tumor response. Hence, a need is created for development new methodologies that can predict the tumor response to radiotherapy at an early stage of the treatment which can lead to tailor-made protocols. To explore the feasibility of prediction of tumor radioresponse, Raman spectra of cervix cancer tissues that were collected before (malignant) and 24 h after patient was treated with 2nd fraction of radiotherapy (RT) were recorded. Data were analyzed by Principal Components Analysis (PCA) and results were correlated with clinical evaluation of radioresponse. Mean Raman spectra of RT tissues corresponding to different levels of tumor response, complete, partial, and no response, showed minute but significant variations. The unsupervised PCA of malignant tissues failed to provide any classification whereas RT spectra gave clear classification between responding (complete and partial response) and nonresponding conditions as well as a tendency of separation among responding conditions. These results were corroborated by supervised classification, by means of discrimination parameters: Mahalanobis distance and spectral residuals. Thus, findings of the study suggest the feasibility of Raman spectroscopic prediction of tumor radioresponse in cervical cancers.     

Inside Cover

Confocal Raman microscopy provides composition and constitution of label-free samples at high spatial-resolution. In this study, the histology of atherosclerotic arteries was measured using a custom-built confocal Raman microscopy that was developed to improve imaging speed, diffraction efficiency, and spectral resolution. Additionally, the machine learning method was used to enhance the accuracy of classification. Results show that each layer of arteries and lipid-rich inflamed plaque can be successfully characterized. Further details can be found in the article by Jingchao Xing, Dong-Ryoung Lee, Jin Won Kim, and Hongki Yoo ( e202200243 ).

     

Photon emission and changes in fluorescent properties of bone after laser irradiation

Laser scalpels used in medical surgery concentrate light energy, heating the tissues. Recently, we reported thermoluminescence emission from laser-treated soft tissues. Here we investigated the thermo-optical effects caused by a laser operating at 808 nm on animal bones (beef ribs) through luminescence and fluorescence imaging, thermal imaging and scanning electron microscopy. Laser-induced artificial lesions emitted luminescence peaking around 650 nm, with a half-life of almost 1 hour. As concerns fluorescence, 24 hours after laser treatment we observed an increase of the emission and a shift from 500 (untreated) to 580 nm (treated). Recrystallization observed by SEM indicates that the temperature in the artificial lesions is over 600°C. We can conclude that laser treatment induces specific luminescent and fluorescent emissions due to heating of the bone and modification of its components. Monitoring these emissions could help prevent tissue overheating and its potential damages during laser-assisted medical procedures.     

Rapid identification of human ovarian cancer in second harmonic generation images using radiomics feature analyses and tree‐based pipeline optimization tool

Ovarian cancer is currently one of the most common cancers of the female reproductive organs, and its mortality rate is the highest among all types of gynecologic cancers. Rapid and accurate classification of ovarian cancer plays an important role in the determination of treatment plans and prognoses. Nevertheless, the most commonly used classification method is based on histopathological specimen examination, which is time‐consuming and labor‐intensive. Thus, in this study, we utilize radiomics feature extraction methods and the automated machine learning tree‐based pipeline optimization tool (TOPT) for analysis of 3D, second harmonic generation images of benign, malignant and normal human ovarian tissues, to develop a high‐efficiency computer‐aided diagnostic model. Area under the receiver operating characteristic curve values of 0.98, 0.96 and 0.94 were obtained, respectively, for the classification of the three tissue types. Furthermore, this approach can be readily applied to other related tissues and diseases, and has great potential for improving the efficiency of medical diagnostic processes.     

Deep learning a boon for biophotonics?

This review covers original articles using deep learning in the biophotonic field published in the last years. In these years deep learning, which is a subset of machine learning mostly based on artificial neural network geometries, was applied to a number of biophotonic tasks and has achieved state‐of‐the‐art performances. Therefore, deep learning in the biophotonic field is rapidly growing and it will be utilized in the next years to obtain real‐time biophotonic decision‐making systems and to analyze biophotonic data in general. In this contribution, we discuss the possibilities of deep learning in the biophotonic field including image classification, segmentation, registration, pseudostaining and resolution enhancement. Additionally, we discuss the potential use of deep learning for spectroscopic data including spectral data preprocessing and spectral classification. We conclude this review by addressing the potential applications and challenges of using deep learning for biophotonic data.     

Histological classification of atherosclerotic arteries using high-speed confocal Raman microscopy with machine learning

Confocal Raman microscopy is a useful tool to observe composition and constitution of label-free samples at high spatial resolution. However, accurate characterization of microstructure of tissue and its application in diagnostic imaging are challenging due to weak Raman scattering signal and complex chemical composition of tissue. We have developed a method to improve imaging speed, diffraction efficiency, and spectral resolution of confocal Raman microscopy. In addition to the novel imaging technique, the machine learning method enables confocal Raman microscopy to visualize accurate histology of tissue sections. Here, we have demonstrated the performance of the proposed method by measuring histological classification of atherosclerotic arteries and compared the histological confocal Raman images with the conventional staining method. Our new confocal Raman microscopy enables us to comprehend the structure and biochemical composition of tissue and diagnose the buildup of atherosclerotic plaques in the arterial wall without labeling.     

In situ measurement of the isoplanatic patch for imaging through intact bone

Wavefront-shaping (WS) enables imaging through scattering tissues like bone, which is important for neuroscience and bone-regeneration research. WS corrects for the optical aberrations at a given depth and field-of-view (FOV) within the sample; the extent of the validity of which is limited to a region known as the isoplanatic patch (IP). Knowing this parameter helps to estimate the number of corrections needed for WS imaging over a given FOV. In this paper, we first present direct transmissive measurement of murine skull IP using digital optical phase conjugation based focusing. Second, we extend our previously reported phase accumulation ray tracing (PART) method to provide in-situ in-silico estimation of IP, called correlative PART (cPART). Our results show an IP range of 1 to 3 μm for mice within an age range of 8 to 14 days old and 1.00 ± 0.25 μm in a 12-week old adult skull. Consistency between the two measurement approaches indicates that cPART can be used to approximate the IP before a WS experiment, which can be used to calculate the number of corrections required within a given field of view.     

Cancer cell‐specific protein delivery by optoporation with laser‐irradiated gold nanorods

The delivery of macromolecules into living cells is challenging since in most cases molecules are endocytosed and remain in the endo‐lysosomal pathway where they are degraded before reaching their target. Here, a method is presented to selectively improve cell membrane permeability by nanosecond laser irradiation of gold nanorods (GNRs) with visible or near‐infrared irradiation in order to deliver proteins across the plasma membrane, avoiding the endo lysosomal pathway. GNRs were labeled with the anti‐EGFR (epidermal growth factor receptor) antibody Erbitux to target human ovarian carcinoma cells OVCAR‐3. Irradiation with nanosecond laser pulses at wavelengths of 532 nm or 730 nm is used for transient permeabilization of the cell membranes. As a result of the irradiation, the uptake of an anti‐Ki‐67 antibody was observed in about 50 % of the cells. The results of fluorescence lifetime imaging show that the GNR detached from the membrane after irradiation.     

Determination of causes of death via spectrochemical analysis of forensic autopsies‐based pulmonary edema fluid samples with deep learning algorithm

This study investigated whether infrared spectroscopy combined with a deep learning algorithm could be a useful tool for determining causes of death by analyzing pulmonary edema fluid from forensic autopsies. A newly designed convolutional neural network‐based deep learning framework, named DeepIR and eight popular machine learning algorithms, were used to construct classifiers. The prediction performances of these classifiers demonstrated that DeepIR outperformed the machine learning algorithms in establishing classifiers to determine the causes of death. Moreover, DeepIR was generally less dependent on preprocessing procedures than were the machine learning algorithms; it provided the validation accuracy with a narrow range from 0.9661 to 0.9856 and the test accuracy ranging from 0.8774 to 0.9167 on the raw pulmonary edema fluid spectral dataset and the nine preprocessing protocol‐based datasets in our study. In conclusion, this study demonstrates that the deep learning‐equipped Fourier transform infrared spectroscopy technique has the potential to be an effective aid for determining causes of death.     

Label-free detection of living cervical cells based on microfluidic device with terahertz spectroscopy

Early diagnosis of cervical cancer is essential for a good prognosis. Terahertz wave detection technology is a nondestructive and label-free physical detection technology, which can detect and monitor the cancer cells in real time, especially for patients with deep or inaccessible tumors. In this study, a single-cell-layer microfluidic device was developed. After replacing the optical clearing agent, the characteristics of H8, HeLa and SiHa cell lines in adherent and suspended states were detected. Additionally, the absorption increased with increasing cell density. For the mixed suspension cell samples, principal component analysis–support vector machine method was used to identify benign and malignant cell component. After living cells formaldehyde, changes in cell membrane permeability were evaluated to identify the cell survival status (i.e., dead or living) based on terahertz spectroscopy amplitude differences. Therefore, extending the terahertz spectrum detection to the molecular level can characterize the life essence of cells and tissues.     

Cross-domain retinopathy classification with optical coherence tomography images via a novel deep domain adaptation method

Deep learning based retinopathy classification with optical coherence tomography (OCT) images has recently attracted great attention. However, existing deep learning methods fail to work well when training and testing datasets are different due to the general issue of domain shift between datasets caused by different collection devices, subjects, imaging parameters, etc. To address this practical and challenging issue, we propose a novel deep domain adaptation (DDA) method to train a model on a labeled dataset and adapt it to an unlabelled dataset (collected under different conditions). It consists of two modules for domain alignment, that is, adversarial learning and entropy minimization. We conduct extensive experiments on three public datasets to evaluate the performance of the proposed method. The results indicate that there are large domain shifts between datasets, resulting a poor performance for conventional deep learning methods. The proposed DDA method can significantly outperform existing methods for retinopathy classification with OCT images. It achieves retinopathy classification accuracies of 0.915, 0.959 and 0.990 under three cross-domain (cross-dataset) scenarios. Moreover, it obtains a comparable performance with human experts on a dataset where no labeled data in this dataset have been used to train the proposed DDA method. We have also visualized the learnt features by using the t-distributed stochastic neighbor embedding (t-SNE) technique. The results demonstrate that the proposed method can learn discriminative features for retinopathy classification.     

Deep learning: step forward to high-resolution in vivo shortwave infrared imaging

Shortwave infrared window (SWIR: 1000–1700 nm) represents a major improvement compared to the NIR-I region (700–900 nm) in terms of temporal and spatial resolutions in depths down to 4 mm. SWIR is a fast and cheap alternative to more precise methods such as X-ray and opto-acoustic imaging. Main obstacles in SWIR imaging are the noise and scattering from tissues and skin that reduce the precision of the method. We demonstrate that the combination of SWIR in vivo imaging in the NIR-IIb region (1500–1700 nm) with advanced deep learning image analysis allows to overcome these obstacles and making a large step forward to high resolution imaging: it allows to precisely segment vessels from tissues and noise, provides morphological structure of the vessels network, with learned pseudo-3D shape, their relative position, dynamic information of blood vascularization in depth in small animals and distinguish the vessels types: artieries and veins. For demonstration we use neural network IterNet that exploits structural redundancy of the blood vessels, which provides a useful analysis tool for raw SWIR images.     

Multimodal imaging for tumour characterization from micro‐ to macroscopic level using a newly developed dorsal chamber designed for long‐term follow‐up

Optical imaging of living animals is a unique method of studying the dynamics of physiological and pathological processes at a subcellular level. One‐shot acquisitions at high resolution can be achieved on exteriorized organs before animal euthanasia. For longitudinal follow‐up, intravital imaging can be used and involves imaging windows implanted in cranial, thoracic or dorsal regions. Several imaging window models exist, but none have proven to be applicable for long‐term monitoring and most biological processes take place over several weeks. Moreover, none are compatible with multiple imaging modalities, meaning that different biological parameters cannot be assessed in an individual animal. We developed a new dorsal chamber that was well tolerated by mice (over several months) and allowed individual and collective cell tracking and behaviour analysis by optical imaging, ultrasound and magnetic resonance tomography. This new model broadens potential applications to areas requiring study of long‐term biological processes, as in cancer research.     

Evaluation of the Arabin cervical pessary for prevention of preterm birth in women with a twin pregnancy and short cervix (STOPPIT-2): An open-label randomised trial and updated meta-analysis

BackgroundPreterm-labour-associated preterm birth is a common cause of perinatal mortality and morbidity in twin pregnancy. We aimed to test the hypothesis that the Arabin pessary would reduce preterm-labour-associated preterm birth by 40% or greater in women with a twin pregnancy and a short cervix.Methods and findingsWe conducted an open-label randomised controlled trial in 57 hospital antenatal clinics in the UK and Europe. From 1 April 2015 to 14 February 2019, 2,228 women with a twin pregnancy underwent cervical length screening between 18 weeks 0 days and 20 weeks 6 days of gestation. In total, 503 women with cervical length ≤ 35 mm were randomly assigned to pessary in addition to standard care (n = 250, mean age 32.4 years, mean cervical length 29 mm, with pessary inserted in 230 women [92.0%]) or standard care alone (n = 253, mean age 32.7 years, mean cervical length 30 mm). The pessary was inserted before 21 completed weeks of gestation and removed at between 35 and 36 weeks or before birth if earlier. The primary obstetric outcome, spontaneous onset of labour and birth before 34 weeks 0 days of gestation, was present in 46/250 (18.4%) in the pessary group compared to 52/253 (20.6%) following standard care alone (adjusted odds ratio [aOR] 0.87 [95% CI 0.55–1.38], p = 0.54). The primary neonatal outcome—a composite of any of stillbirth, neonatal death, periventricular leukomalacia, early respiratory morbidity, intraventricular haemorrhage, necrotising enterocolitis, or proven sepsis, from birth to 28 days after the expected date of delivery—was present in 67/500 infants (13.4%) in the pessary group compared to 76/506 (15.0%) following standard care alone (aOR 0.86 [95% CI 0.54–1.36], p = 0.50). The positive and negative likelihood ratios of a short cervix (≤35 mm) to predict preterm birth before 34 weeks were 2.14 and 0.83, respectively. A meta-analysis of data from existing publications (4 studies, 313 women) and from STOPPIT-2 indicated that a cervical pessary does not reduce preterm birth before 34 weeks in women with a short cervix (risk ratio 0.74 [95% CI 0.50–1.11], p = 0.15). No women died in either arm of the study; 4.4% of babies in the Arabin pessary group and 5.5% of babies in the standard treatment group died in utero or in the neonatal period (p = 0.53). Study limitations include lack of power to exclude a smaller than 40% reduction in preterm labour associated preterm birth, and to be conclusive about subgroup analyses.ConclusionsThese results led us to reject our hypothesis that the Arabin pessary would reduce the risk of the primary outcome by 40%. Smaller treatment effects cannot be ruled out.Trial registrationISRCTN Registry ISRCTN 02235181.ClinicalTrials.gov {"type":"clinical-trial","attrs":{"text":"NCT02235181","term_id":"NCT02235181"}}NCT02235181.

Jane Norman and co-workers report an evaluation of a cervical pessary for prevention of preterm birth in women with a short cervix bearing twins.     

Diagnosing colorectal abnormalities using scattering coefficient maps acquired from optical coherence tomography

Optical coherence tomography (OCT) has shown potential in differentiating normal colonic mucosa from neoplasia. In this study of 33 fresh human colon specimens, we report the first use of texture features and computer vision-based imaging features acquired from en face scattering coefficient maps to characterize colorectal tissue. En face scattering coefficient maps were generated automatically using a new fast integral imaging algorithm. From these maps, a gray-level cooccurrence matrix algorithm was used to extract texture features, and a scale-invariant feature transform algorithm was used to derive novel computer vision-based features. In total, 25 features were obtained, and the importance of each feature in diagnosis was evaluated using a random forest model. Two classifiers were assessed on two different classification tasks. A support vector machine model was found to be optimal for distinguishing normal from abnormal tissue, with 94.7% sensitivity and 94.0% specificity, while a random forest model performed optimally in further differentiating abnormal tissues (i.e., cancerous tissue and adenomatous polyp) with 86.9% sensitivity and 85.0% specificity. These results demonstrated the potential of using OCT to aid the diagnosis of human colorectal disease.     

Optimizing the classification of biological tissues using machine learning models based on polarized data

Polarimetric data is nowadays used to build recognition models for the characterization of organic tissues or the early detection of some diseases. Different Mueller matrix-derived polarimetric observables, which allow a physical interpretation of a specific characteristic of samples, are proposed in literature to feed the required recognition algorithms. However, they are obtained through mathematical transformations of the Mueller matrix and this process may loss relevant sample information in search of physical interpretation. In this work, we present a thorough comparative between 12 classification models based on different polarimetric datasets to find the ideal polarimetric framework to construct tissues classification models. The study is conducted on the experimental Mueller matrices images measured on different tissues: muscle, tendon, myotendinous junction and bone; from a collection of 165 ex-vivo chicken thighs. Three polarimetric datasets are analyzed: (A) a selection of most representative metrics presented in literature; (B) Mueller matrix elements; and (C) the combination of (A) and (B) sets. Results highlight the importance of using raw Mueller matrix elements for the design of classification models.