Automated system classifies skin cancers <\/p>\n
Skin cancer, the most common human malignancy, is usually diagnosed visually and then confirmed with follow-up biopsies and histological tests. Automated classification of skin lesions is desirable but challenging because such lesions vary greatly in appearance. Now, researchers from Stanford University have devised a deep-learning algorithm that can classify skin cancers from images. They trained the algorithm using a dataset of 129,450 clinical images representing more than 2000 different skin diseases. In tests on clinical images, the algorithm could diagnose the most common and the most deadly types of skin cancer malignant carcinomas and melanomas, respectively with equivalent performance to a group of 21 board-certified dermatologists (Nature 542 115). <\/p>\n
\"We made a very powerful machine learning algorithm that learns from data,\" said Andre Esteva, co-lead author of the paper and a graduate student in the Thrun lab. \"Instead of writing into computer code exactly what to look for, you let the algorithm figure it out.\" The authors note that the system has yet to be validated in a real-world clinical setting, but has extensive potential to affect primary care. They also hope to make the algorithm smartphone compatible in the near future. \"My main eureka moment was when I realized just how ubiquitous smartphones will be,\" added Esteva. \"Everyone will have a supercomputer in their pockets with a number of sensors in it, including a camera. What if we could use it to visually screen for skin cancer? Or other ailments?\" <\/p>\n
Researchers at the University of Michigan Medical School have designed a portable cancer diagnostic system that enables faster and more accurate diagnosis of brain tumours in the operating room. Typically, after removing the tumour, the surgeon must wait 30 to 40 minutes while the tissue is sent to a pathology lab for processing, sectioning, staining, mounting and interpretation. This can delay decision-making in the operating room, while tissue processing can introduce artefacts. Instead, the Michigan researchers have developed a stimulated Raman histology (SRH) system that can provide fast analysis of fresh brain tumour samples in the operating room, with no sample processing or staining required (Nature Biomedical Engineering 1 0027). <\/p>\n
SRH is based on stimulated Raman scattering microscopy, using a fibre-laser-based microscope. The technology produces images that are virtually coloured to highlight cellular and architectural features and are almost indistinguishable from traditionally stained samples. The researchers imaged tissue from 101 neurosurgical patients using the new approach and conventional methods. Both produced accurate results but SRH was much faster. Neuropathologists given 30 samples, processed via SRH or traditional methods, were equally likely to make a correct diagnosis with either sample. The team has also built a machine learning process that could predict brain tumour subtype with 90% accuracy. \"By achieving excellent image quality in fresh tissues, we're able to make a diagnosis during surgery,\" said first author Daniel Orringer. \"Our technique may disrupt the intraoperative diagnosis process in a great way, reducing it from a 30-minute process to about three minutes. Initially, we developed this technology as a means of helping surgeons detect microscopic tumour, but we found the technology was capable of much more than guiding surgery.\" <\/p>\n
A research team headed up at the Center for Nanomedicine in the Republic of Korea has developed the Nano MRI Lamp a platform based on an MRI contrast that only \"switches on\" in the presence of the targeted disease. The Nano MRI Lamp technology combines two magnetic materials: a superparamagnetic quencher (magnetic nanoparticle) and a paramagnetic enhancer (MRI contrast agent). When the two materials are separated by more than 7nm, the MRI signal is on, whereas when they are placed closer than 7nm, the signal is switched off. The researchers named this approach magnetic resonance tuning (Nature Materials doi: 10.1038\/nmat4846). <\/p>\n
The team tested the Nano MRI Lamp's performance by detecting the presence of MMP-2, an enzyme that can induce tumour metastasis, in mice with cancer. They connected the two magnetic materials with a linker, bringing them close together and switching the MRI signal off. In the presence of cancer, the MMP-2 cleaves this linker, separating the materials and switching the MRI signal on. The resulting MR image thus indicated the location of the tumour, with the signal brightness correlated with MMP-2 concentration in the cancerous tissue. \"The current contrast agent is like using a flashlight during a sunny day: its effect is limited. Instead, this new technology is like using a flash light at night and therefore more useful,\" explained team leader Jinwoo Cheon. <\/p>\n
The first-in-human application of a PET radiotracer designed to identify both early and metastatic prostate cancer has been reported by a USChina research team. The new tracer is a Ga-68-labelled peptide BBN-RGD agent that targets both gastrin-releasing peptide receptor and integrin v3, both of which are overexpressed in prostate cancer cells. The study included 13 patients with prostate cancer (four newly diagnosed and nine post-therapy) and five healthy volunteers. PET\/CT using Ga-68-BBN-RGD detected 20 bone lesions in seven patients, either with primary prostate cancer or after radical prostatectomy. No adverse side effects were found during the procedure and two-week follow-up, demonstrating the safety of the radiotracer (J. Nucl. Med. 58 228). <\/p>\n
\"Compounds capable of targeting more than one biomarker have the ability of binding to both early and metastatic stages of prostate cancer, creating the possibility for a more prompt and accurate diagnostic profile for both primary and the metastatic tumours,\" explained senior investigator Xiaoyuan Chen, from the Laboratory of Molecular Imaging and Nanomedicine at NIBIB. Looking ahead, Chen says that Ga-68-BBN-RGD could play a role in staging and detecting prostate cancer and provide guidance for internal radiation therapy, using the same peptide labelled with therapeutic radionuclides. He points out that larger-scale clinical investigations are warranted. <\/p>\n
MSOM offers 3D in vivo skin mapping Raman imaging steps closer to the clinic Multifunctional bubbles image and treat PET helps quantify bone metastases response <\/p>\n
<\/p>\n
View post:<\/p>\n
Research briefs: diagnostic imaging - Medical Physics Web (subscription)<\/a><\/p>\n","protected":false},"excerpt":{"rendered":" Automated system classifies skin cancers Skin cancer, the most common human malignancy, is usually diagnosed visually and then confirmed with follow-up biopsies and histological tests. Automated classification of skin lesions is desirable but challenging because such lesions vary greatly in appearance. Continue reading