Deep learning has revolutionized the detection of diseases and is helping the healthcare sector break barriers in terms of accuracy and robustness to achieve efficient and robust computer-aided diagnostic systems. The application of deep learning techniques empowers automated AI-based utilities requiring minimal human supervision to perform any task related to medical diagnosis of fractures, tumors, and internal hemorrhage; preoperative planning; intra-operative guidance, etc. But deep learning faces some major threats to the flourishing healthcare domain. This paper traverses the major challenges that the deep learning community of researchers and engineers faces, particularly in medical image diagnosis, like the unavailability of balanced annotated medical image data, adversarial attacks faced by deep neural networks and architectures due to noisy medical image data, a lack of trustability among users and patients, and ethical and privacy issues related to medical data. This study explores the possibilities of AI autonomy in healthcare by overcoming the concerns about trust that society has in autonomous intelligent systems.

In this paper, a blind medical image watermarking scheme based on Fast Discrete Curvelet Transform (FDCuT) and Discrete Cosine Transform (DCT) is proposed. FDCuT is applied on the medical image to get different frequency coefficients of its curvelet decomposition. On the high frequency curvelet coefficients of the medical image, block wise DCT is applied to get different frequency coefficients. Then mid band frequency coefficients of the medical image are modified by White Gaussian Noise (WGN) sequences according to watermark bit to get watermarked medical image. At extraction end, blind recovery of watermark data is performed by correlation of WGN sequences. The proposed scheme is tested for its effectiveness on various types of medical images such as X-ray, Ultrasound (US), Magnetic Resonant Imaging (MRI) and Computerized Tomography (CT). Result analysis shows that imperceptibility of watermarked medical image is better as PSNR is above 45 dB for all types of the medical images. In addition, the robustness of the scheme is better than an existing scheme for a similar set of medical images in terms of Normalized Correlation (NC). Experimental results show that scheme is robust to geometric attacks, signal processing attacks and JPEG compression attacks. An analysis is also carried out to verify the performance of the proposed scheme to support binary watermarks with different details in it: text and logos. Moreover, the proposed scheme resulted in zero false positive rate when tested on 100 non-watermarked images

In this paper, a hybrid approach based on cryptography and steganography is proposed for the security of medical image over an open communication channel. In this approach, medical image information is embedded using discrete cosine transform (DCT)-singular value decomposition (SVD)-based embedding process. The medical image is encrypted using CS encryption before embedding into the standard image. This encrypted medical image is inserted into the singular values of DCT coefficients of the standard image to get the stego image. The experimental results show that the encrypted medical image is successfully extracted from the stego image under various image processing attacks at recovering side. The result analysis also shows improved imperceptibility of the stego image with a peak signal-to-noise ratio value above 60 dB for all types of medical images under consideration. Furthermore, the computation time of proposed approach is less than that of existing approaches.

Attendance recording of a student in an academic organization plays a vital role in judging students performance. As manual labor involved in this process is time consuming, an automated Attendance Management System (AMS) based on face detection and face recognition techniques is proposed in this paper. The system employs modified Viola-Jones algorithm for face detection, and alignment- free partial face recognition algorithm for face recognition. After successful recognition of a student, the system automatically updates the attendance in the excel sheet. The proposed system improves the performance of existing attendance management systems by eliminating manual calling, marking and entry of attendance in institutional websites.