XRelevanceCAM: towards explainable tissue characterization with improved localisation of pathological structures in probe-based confocal laser endomicroscopy.

PURPOSE: Probe-based confocal laser endomicroscopy (pCLE) enables intraoperative tissue characterization with improved resection rates of brain tumours. Although a plethora of deep learning models have been developed for automating tissue characterization, their lack of transparency is a concern. To tackle this issue, techniques like Class Activation Map (CAM) and its variations highlight image regions related to model decisions. However, they often fall short of providing human-interpretable visual explanations for surgical decision support, primarily due to the shattered gradient problem or insufficient theoretical underpinning. METHODS: In this paper, we introduce XRelevanceCAM, an explanation method rooted in a better backpropagation approach, incorporating sensitivity and conservation axioms. This enhanced method offers greater theoretical foundation and effectively mitigates the shattered gradient issue when compared to other CAM variants. RESULTS: Qualitative and quantitative evaluations are based on ex vivo pCLE data of brain tumours. XRelevanceCAM effectively highlights clinically relevant areas that characterize the tissue type. Specifically, it yields a remarkable 56% improvement over our closest baseline, RelevanceCAM, in the network's shallowest layer as measured by the mean Intersection over Union (mIoU) metric based on ground-truth annotations (from 18 to 28.07%). Furthermore, a 6% improvement in mIoU is observed when generating the final saliency map from all network layers. CONCLUSION: We introduce a new CAM variation, XRelevanceCAM, for precise identification of clinically important structures in pCLE data. This can aid introperative decision support in brain tumour resection surgery, as validated in our performance study.

Development, Implementation and Application of Confocal Laser Endomicroscopy in Brain, Head and Neck Surgery-A Review.

When we talk about visualization methods in surgery, it is important to mention that the diagnosis of tumors and how we define tumor borders intraoperatively in a correct way are two main things that would not be possible to achieve without this grand variety of visualization methods we have at our disposal nowadays. In addition, histopathology also plays a very important role, and its importance cannot be neglected either. Some biopsy specimens, e.g., frozen sections, are examined by a histopathologist and lead to tumor diagnosis and the definition of its borders. Furthermore, surgical resection is a very important point when it comes to prognosis and life survival. Confocal laser endomicroscopy (CLE) is an imaging technique that provides microscopic information on the tissue in real time. CLE of disorders, such as head, neck and brain tumors, has only recently been suggested to contribute to both immediate tumor characterization and detection. It can be used as an additional tool for surgical biopsies during biopsy or surgical procedures and for inspection of resection margins during surgery. In this review, we analyze the development, implementation, advantages and disadvantages as well as the future directions of this technique in neurosurgical and otorhinolaryngological disciplines.

Confocal Laser Endomicroscopy in Oncological Surgery.

The therapy of choice in the treatment of abnormalities in the human body, is to attempt a personalized diagnosis and with minimal invasiveness, ideally resulting in total resection (surgery) or turning off (intervention) of the pathology with preservation of normal functional tissue, followed by additional treatments, e [...].

Compact and contactless reflectance confocal microscope for neurosurgery.

Visual guidance at the cellular level during neurosurgical procedures is essential for complete tumour resection. We present a compact reflectance confocal microscope with a 20 mm working distance that provided <1.2 µm spatial resolution over a 600 µm × 600 µm field of view in the near-infrared region. A physical footprint of 200 mm × 550 mm was achieved using only standard off-the-shelf components. Theoretical performance of the optical design was first evaluated via commercial Zemax software. Then three specimens from rodents: fixed brain, frozen calvaria and live hippocampal slices, were used to experimentally assess system capability and robustness. Results show great potential for the proposed system to be translated into use as a next generation label-free and contactless neurosurgical microscope.

5-Aminolevulinic Acid Multispectral Imaging for the Fluorescence-Guided Resection of Brain Tumors: A Prospective Observational Study.

Fluorescence-guided surgery with five-aminolevulinic acid (5-ALA) is the state-of-the-art treatment of high-grade gliomas. However, intraoperative visualization of 5-ALA under blue light remains challenging, especially when blood covers the surgical field and thereby fluorescence. To overcome this problem and combine the brightness of visible light with the information delivered with fluorescence, we implemented multispectral fluorescence (MFL) in a surgical microscope, a technique that is able to project both information in real-time. We prospectively examined 25 patients with brain tumors. One patient was operated on two different lesions in the same setting. The tumors comprised: six glioblastomas, four anaplastic astrocytomas, one anaplastic oligodendroglioma, two meningiomas, 11 metastatic tumors, one acoustic neuroma, and one ependymoma. The MFL technique with a real-time overlay of fluorescence and white light was compared intraoperatively to the classic blue filter. All lesions were clearly visible and highlighted from the surrounding tissue. The pseudocolor we chose was green, representing fluorescence, with the surrounding brain tissue remaining in its original color. When blood was covering the surgical field, orientation was easy to maintain. The MFL technique opens the way for precise and clear visualization of fluorescence in real-time under white light. It can be easily used for the resection of all tumors accumulating 5-ALA. Drawbacks of classic PpIX fluorescence such as hidden fluorescence, intraoperative changes could be overcome with the presence of additional white light in MFL technique.

Real-Time Overlapping of Indocyanine Green-Video Angiography With White Light Imaging for Vascular Neurosurgery: Technique, Implementation, and Clinical Experience.

BACKGROUND: Fluorescent-guided techniques in vascular neurosurgery can be demonstrated via black and white indocyanine green videoangiography (ICG-VA). Multispectral imaging (MFL) is a new method, which overlaps fluorescence with the white light and provides a fluorescent white light augmented reality image to the surgeon. OBJECTIVE: To investigate (a) whether MFL can enhance the visualization of the blood-flow with simultaneous visualization of the anatomic structures and (b) if MFL can ergonomically improve the microvascular surgical treatment compared to ICG-VA. METHODS: A digital imaging of the blood flow after intravenous injection of ICG on 7 pigs was performed in real time under white light, standard fluorescence, and MFL. The blood flow was interrupted with a surgical clip, demonstrating the blockage of the blood flow. We prospectively included 30 patients with vascular deformities. The vasculature was visualized on the microscope's monitor and through the microscope's eyepiece. RESULTS: In the animal experiment, the visualization of the anatomy and the blood flow under MFL produced high resolution images. The occlusion of blood vessels demonstrated sufficiently the blockage of tissue perfusion and its reperfusion after clip removal. During all 30 surgical cases, the MFL technique and the direct delivery of the pseudo-colored image through the eyepiece allowed for enhanced anatomic and dynamic data. CONCLUSION: MFL was shown to be superior to the classic ICG-VA, delivering enhanced data and notably improving the workflow due to the simultaneous and precise white light visualization of the blood flow and the surrounding anatomic structures.

Confocal-Assisted Multispectral Fluorescent Microscopy for Brain Tumor Surgery.

Optimal surgical therapy for brain tumors is the combination of complete resection with minimal invasion and damage to the adjacent normal tissue. To achieve this goal, we need advanced imaging techniques on a scale from macro- to microscopic resolution. In the last decade, the development of fluorescence-guided surgery has been the most influential breakthrough, marginally improving outcomes in brain tumor surgery. Multispectral fluorescence microscopy (MFL) is a novel imaging technique that allows the overlapping of a fluorescent image and a white light image in real-time, with delivery of the merged image to the surgeon through the eyepieces of a surgical microscope. MFL permits the detection and characterization of brain tumors using fluorescent molecular markers such as 5-aminolevulinic acid (5-ALA) or indocyanine green (ICG), while simultaneously obtaining high definition white light images to create a pseudo-colored composite image in real-time. Limitations associated with the use of MFL include decreased light imaging intensity and decreased levels of magnification that may compromise maximal tumor resection on a cellular scale. Confocal laser endomicroscopy (CLE) is another novel advanced imaging technique that is based on miniaturization of the microscope imaging head in order to provide the possibility of in vivo microscopy at the cellular level. Clear visualization of the cellular cytoarchitecture can be achieved with 400-fold-1,000-fold magnification. CLE allows on the one hand the intra-operative detection and differentiation of single tumor cells (without the need for intra-operative histologic analysis of biopsy specimens) as well as the definition of borders between tumor and normal tissue at a cellular level, dramatically improving the accuracy of surgical resection. The application and implementation of CLE-assisted surgery in surgical oncology increases not only the number of options for real-time diagnostic imaging, but also the therapeutic options by extending the resection borders of cancer at a cellular level and, more importantly, by protecting the functionality of normal tissue in the adjacent areas of the human brain. In this article, we describe our experience using these new techniques of confocal-assisted fluorescent surgery including analysis on the technology, usability, indications, limitations, and further developments.

Context aware decision support in neurosurgical oncology based on an efficient classification of endomicroscopic data.

PURPOSE: Probe-based confocal laser endomicroscopy (pCLE) enables in vivo, in situ tissue characterisation without changes in the surgical setting and simplifies the oncological surgical workflow. The potential of this technique in identifying residual cancer tissue and improving resection rates of brain tumours has been recently verified in pilot studies. The interpretation of endomicroscopic information is challenging, particularly for surgeons who do not themselves routinely review histopathology. Also, the diagnosis can be examiner-dependent, leading to considerable inter-observer variability. Therefore, automatic tissue characterisation with pCLE would support the surgeon in establishing diagnosis as well as guide robot-assisted intervention procedures. METHODS: The aim of this work is to propose a deep learning-based framework for brain tissue characterisation for context aware diagnosis support in neurosurgical oncology. An efficient representation of the context information of pCLE data is presented by exploring state-of-the-art CNN models with different tuning configurations. A novel video classification framework based on the combination of convolutional layers with long-range temporal recursion has been proposed to estimate the probability of each tumour class. The video classification accuracy is compared for different network architectures and data representation and video segmentation methods. RESULTS: We demonstrate the application of the proposed deep learning framework to classify Glioblastoma and Meningioma brain tumours based on endomicroscopic data. Results show significant improvement of our proposed image classification framework over state-of-the-art feature-based methods. The use of video data further improves the classification performance, achieving accuracy equal to 99.49%. CONCLUSIONS: This work demonstrates that deep learning can provide an efficient representation of pCLE data and accurately classify Glioblastoma and Meningioma tumours. The performance evaluation analysis shows the potential clinical value of the technique.

Presurgical mapping of basal cell carcinoma or squamous cell carcinoma by confocal laser endomicroscopy compared to traditional micrographic surgery: a single-centre prospective feasibility study.

BACKGROUND: At present, no ideal diagnostic tools exist in the market to excise cancer tissue with the required safety margins and to achieve optimal aesthetic results using tissue-conserving techniques. OBJECTIVES: In this prospective study, confocal laser endomicroscopy (CLE) and the traditional gold standard of magnifying glasses (MG) were compared regarding the boundaries of in vivo basal cell carcinoma and squamous cell carcinoma. MATERIALS & METHODS: Tumour diameters defined by both methods were measured and compared with those determined by histopathological examination. Nineteen patients were included in the study. RESULTS: The CLE technique was found to be superior to excisional margins based on MG only. Re-excision was required in 68% of the cases following excision based on MG evaluation, but only in 27% of the cases for whom excision margins were based on CLE. CONCLUSION: Our results are promising regarding the distinction between tumour and healthy surrounding tissue, and indicate that presurgical mapping of basal cell carcinoma and squamous cell carcinoma is possible. The tool itself should be developed further with special attention to early detection of skin cancer.

Noninvasive histological imaging of head and neck squamous cell carcinomas using confocal laser endomicroscopy.

Confocal laser endomicroscopy (CLE) is an imaging technique that uses miniaturized fiberoptic probes to allow real-time histological imaging of human tissue. An application of CLE in otorhinolaryngology has hardly been investigated so far. In our study, we analyzed the applicability of CLE to visualize cancerous and healthy tissue of the head and neck region. Formalin-fixed tissue specimens from 135 head and neck squamous cell carcinoma (HNSCC) patients and 50 healthy controls were investigated using CLE with and without topical application of acriflavine. Four head and neck surgeons, four pathologists, and four laymen evaluated the CLE images of the HNSCC cases regarding the tumor localization and its border to healthy tissue. The tumor localization and the tumor border were correctly identified in 97 % by the pathologists, 85 % by the head and neck surgeons, and 70 % by the laymen. The main difference in evaluation results was seen in the correct identification of the tumor site (p < 0.05), while there was no significant difference in the identification of the tumor border. CLE is a valuable tool for real-time histological imaging of HNSCCs. It can help to visualize the tumor border and, thereby, facilitate a more precise tumor surgery.

Automatic Tissue Differentiation Based on Confocal Endomicroscopic Images for Intraoperative Guidance in Neurosurgery.

Diagnosis of tumor and definition of tumor borders intraoperatively using fast histopathology is often not sufficiently informative primarily due to tissue architecture alteration during sample preparation step. Confocal laser microscopy (CLE) provides microscopic information of tissue in real-time on cellular and subcellular levels, where tissue characterization is possible. One major challenge is to categorize these images reliably during the surgery as quickly as possible. To address this, we propose an automated tissue differentiation algorithm based on the machine learning concept. During a training phase, a large number of image frames with known tissue types are analyzed and the most discriminant image-based signatures for various tissue types are identified. During the procedure, the algorithm uses the learnt image features to assign a proper tissue type to the acquired image frame. We have verified this method on the example of two types of brain tumors: glioblastoma and meningioma. The algorithm was trained using 117 image sequences containing over 27 thousand images captured from more than 20 patients. We achieved an average cross validation accuracy of better than 83%. We believe this algorithm could be a useful component to an intraoperative pathology system for guiding the resection procedure based on cellular level information.

Endoscope-assisted keyhole surgery via an eyebrow incision for removal of large meningiomas of the anterior and middle cranial fossa.

BACKGROUND: Conventional open surgery of large meningiomas has proven to be challenging even in experienced hands. Intense retraction and dissection around neurovascular structures increase morbidity and mortality. In the present study, we retrospectively analyzed the surgical technique, and outcome in 40 patients with large anterior cranial fossa meningiomas extending to the middle fossa. All patients were approached via a supraorbital mini craniotomy. METHODS: It is a retrospective study of 40 patients (12 males, 28 females) who underwent surgery for large anterior cranial fossa meningiomas (diameter >5 cm) extending to the middle fossa in four different neurosurgical centers within 6 years. Depending on the localization of the tumor, the skin incision was between 2.5 and 3 cm long and was made without shaving the patient's eyebrow hair. Subsequently, a keyhole craniotomy was performed of approximately 0.8×1.2-1.4 cm in diameter. Preoperative and postoperative clinical and radiological data were analyzed and discussed. RESULTS: Headache and psycho-organic syndrome were the most common presenting symptom in all patients. Presenting symptoms were associated with psychological changes in 23 cases, visual impairment in 19 patients, and anosmia in 17 patients. In overall, 36 of 40 patients (90%) showed a good outcome and returned at long-term follow-up to their previous occupations. The elderly patients returned to their daily routine. CONCLUSION: With the appropriate keyhole approach as a refinement of the classic keyhole craniotomy to a smaller key"burr"hole, and with use of modern and new designed equipment, it is possible to perform complete resection of large anterior and middle fossa meningiomas with the same safety, efficiency and with less complication rates as described in the literature for large meningiomas even performed with classic keyhole craniotomies.

Visualization of small veins with susceptibility-weighted imaging for stereotactic trajectory planning in deep brain stimulation.

Intracerebral hemorrhage (ICH) is the most significant complication of Deep Brain Stimulation (DBS). To prevent ICH, stereotactic contrast enhanced T1-weighted images are used to visualize vessels as source of hemorrhage. Susceptibility-Weighted Imaging (SWI) is an MRI sequence with improved visualization of susceptibility differences between tissues, particularly sensitive for brain veins. The aim of this prospective study was to analyze the utility of SWI compared to contrast enhanced stereotactic T1-weighted images for trajectory planning of DBS. Preoperative SWI was performed in 33 patients undergoing DBS and was compared to the T1-weighted images. Vessels identified only with SWI in relation to the bilateral planned trajectory were analyzed. In all patients vessels were depicted on SWI only within the planned trajectory (range 1-4 vessels, for each trajectory, mean: 2.4). In 6 patients vessels were identified on SWI adjacent to the target (up to 5mm distal from target). In 11 patients SWI visualized additional cortical veins adjacent to the entry point of the trajectory. The apparent diameter of these vessels ranged between 0.8 and 2.1mm (mean: 1.2mm). Postoperative MRI was compared with preoperative SWI and revealed in two patients small (<3 mm) T2 hyperintense lesions along electrodes without correlation with visualized veins. SWI facilitates the visualization of small veins superior to T1-weighted images. However, cerebral veins within the trajectory were not found to be a significant source of ICH after DBS. Potential sources of ICH are mesencephal veins at the endpoint of electrodes which can cause fatal hemorrhage and are visualized with SWI reliably.

Endoscopic endonasal transsphenoidal approach for resection of a coexistent pituitary macroadenoma and a tuberculum sellae meningioma.

The coexistence of a pituitary macroadenoma and a tuberculum sellae meningioma is very rare. This article demonstrates the surgical technique of the simultaneous resection of a pituitary macroadenoma and a tuberculum sellae meningioma using an endoscopic, endonasal, biportal, transsphenoidal approach. A 36-year-old woman presented with frontal headache and extended visual field loss of the right eye. She underwent cranial magnetic resonance imaging (MRI) revealing a 2 × 2 × 2.5 mm contrast-enhancing intrasellar and suprasellar lesion with compression of the optic chiasma. The coexistence of a pituitary macroadenoma and meningioma was suggested. A biportal endoscopic endonasal transsphenoidal approach was performed to remove both lesions. The histological results confirmed the coexistence of the pituitary macroadenoma and meningioma, World Health Organization (WHO) grade I. The endoscopic, endonasal, transsphenoidal approach is a safe and reliable minimal invasive surgical alternative for resection of the intra-, supra- and parasellar lesions, avoiding additional craniotomy.

Impedance spectroscopy--an outstanding method for label-free and real-time discrimination between brain and tumor tissue in vivo.

Until today, brain tumors especially glioblastoma are difficult to treat and therefore, results in a poor survival rate of 0-14% over five years. To overcome this problem, the development of novel therapeutics as well as optimization of neurosurgical procedures to remove the tumor tissue are subject of intensive research. The main problem of the tumor excision, as the primary clinical intervention is the diffuse infiltration of the tumor cells in unaltered brain tissue that complicates the complete removal of residual tumor cells. In this context, we are developing novel approaches for the label-free discrimination between tumor tissue and unaltered brain tissue in real-time during the surgical process. Using our impedance spectroscopy-based measurement system in combination with flexible microelectrode arrays we could successfully demonstrate the discrimination between a C6-glioma and unaltered brain tissue in an in vivo rat model. The analysis of the impedance spectra revealed specific impedance spectrum shape characteristics of physiologic neuronal tissue in the frequency range of 10-500 kHz that were significantly different from the tumor tissue. Moreover, we used an adapted equivalent circuit model to get a deeper understanding for the nature of the observed effects. The impedimetric label-free and real-time discrimination of tumor from unaltered brain tissue offers the possibility for the implementation in surgical instruments to support surgeons to decide, which tissue areas should be removed and which should be remained.

Optimal invasive key-hole neurosurgery with a miniaturized 3D chip on the tip: Microendoscopic device.

OBJECTIVE: The goal of the performed study was to evaluate the possibility of a three-dimensional endoscope to become a combined microscope-endoscope device in one. We analyzed the ergonomy of the device, the implementation into the surgical workflow, the image quality, and the future perspectives such devices could have for the next generation of neurosurgeons. MATERIALS AND METHODS: Within 6 months, 22 patients (10 male, 12 female, 20-65 age) underwent surgery in neuroaxis using the new 3D-microendoscope (ME). The new 3D-ME has (a) the ability to visualize the surgical field from out- to inside with all advantages offered by a microscope, and in the same moment, (b) its design is like a small diameter endoscope that allows stereoscopic views extracorporal, intracorporal, and panoramic "para-side" of the lesion. RESULTS: In general, transcranial 3D-"microendoscopy" was performed in all patients with high-resolution 3D quality. No severe complications were observed intra- or postoperatively. With the addition of depth perception, the anatomic structures were well seen and observed. CONCLUSION: The 3D-microendoscopy is a very promising surgical concept associated with new technological developments. The surgeon is able to switch to a modern visualization instrument reaching the most optimal surgical approach without compromising safety, effectiveness, and visual information.

Sub-cellular tumor identification and markerless differentiation in the rat brain in vivo by multiphoton microscopy.

OBJECTIVE/BACKGROUND: Aim of the current study was to localize and differentiate between tumor (glioma) and healthy tissue in rat brains on a cellular level. Near-infrared multiphoton microscopy takes advantage of the simultaneous absorption of two or more photons to analyze various materials such as cell and tissue components via the observation of endogenous fluorophores such as NAD(P)H, FAD, porphyrins, melanin, elastin, and collagen, with a very high resolution, without inducing the problems of photo-bleaching on out-of-focus areas. METHODS: In vitro and in vivo studies on healthy rat brains as well as C6 glioma cell line allografts have been performed. Near-infrared laser pulses (λ = 690-1060 nm, τ ~140 fs) generated by an ultrafast Ti:Sapphire tunable laser system (Chameleon, Coherent GmbH, Santa Clara, CA) were coupled into a laser scanning microscope (LSM 510 META, Carl Zeiss, Germany) to observe high quality images. RESULTS: Several image acquisitions have been performed by varying the zoom scale of the multiphoton microscope, image acquisition time and the wavelength (765, 840 nm) to detect various tissue components. With a penetration depth of ~200 µm in vitro and about 30-60 µm in vivo into the brain tissue it was possible to differentiate between tumor and healthy brain tissue even through thin layers of blood. CONCLUSION: Near-infrared multiphoton microscopy allows the observation and possibly differentiation between tumor (glioma) and healthy tissue in rat brains on a cellular level. Our findings suggest that a further miniaturization of this technology might be very useful for scientific and clinical applications in neurosurgery.

Confocal laser endomicroscopy for diagnosis and histomorphologic imaging of brain tumors in vivo.

Early detection and evaluation of brain tumors during surgery is crucial for accurate resection. Currently cryosections during surgery are regularly performed. Confocal laser endomicroscopy (CLE) is a novel technique permitting in vivo histologic imaging with miniaturized endoscopic probes at excellent resolution. Aim of the current study was to evaluate CLE for in vivo diagnosis in different types and models of intracranial neoplasia. In vivo histomorphology of healthy brains and two different C6 glioma cell line allografts was evaluated in rats. One cell line expressed EYFP, the other cell line was used for staining with fluorescent dyes (fluorescein, acriflavine, FITC-dextran and Indocyanine green). To evaluate future application in patients, fresh surgical resection specimen of human intracranial tumors (n = 15) were examined (glioblastoma multiforme, meningioma, craniopharyngioma, acoustic neurinoma, brain metastasis, medulloblastoma, epidermoid tumor). Healthy brain tissue adjacent to the samples served as control. CLE yielded high-quality histomorphology of normal brain tissue and tumors. Different fluorescent agents revealed distinct aspects of tissue and cell structure (nuclear pattern, axonal pathways, hemorrhages). CLE discrimination of neoplastic from healthy brain tissue was easy to perform based on tissue and cellular architecture and resemblance with histopathology was excellent. Confocal laser endomicroscopy allows immediate in vivo imaging of normal and neoplastic brain tissue at high resolution. The technology might be transferred to scientific and clinical application in neurosurgery and neuropathology. It may become helpful to screen for tumor free margins and to improve the surgical resection of malignant brain tumors, and opens the door to in vivo molecular imaging of tumors and other neurologic disorders.

Surgical complications after endoscopic transsphenoidal pituitary surgery.

Between January 2004 and June 2007 we conducted a retrospective analysis to assess post-operative complications related to endoscopic pituitary surgery in a series of 150 patients. Patients were treated with an endoscopic endonasal transsphenoidal approach to the sellar region for removal of pathological sellar and suprasellar lesions. We analysed the complications in groups according to the anatomical structures of the approach and the functional systems of the pituitary gland (anterior and posterior endocrine systems), and compared them to a large historical series using the traditional microsurgical transsphenoidal approach. Overall, we observed a decreased incidence of complications with respect to the surgical trauma, the functionality of the pituitary gland and post-operative patient comfort. We believe that the reduction of the complication rate observed in this study was mainly due to the wide structural overview offered by the endoscope as well as the anatomically direct, and therefore minimally invasive, character of the procedure. Successful endoscopic pituitary surgery requires extensive training in the use of an endoscope and careful planning of the surgery. Furthermore, close cooperation between a multidisciplinary team consisting of endocrinologists, neurosurgeons, ear, nose and throat surgeons, radiologists, and radiation oncologists is of utmost importance.