Using multimodal biometrics, data hiding and encryption for secure healthcare imaging system

In  this  digital  era,  images  are  the  most  vital  in-formation  carrier  used  for  healthcare  communication  and  en-tertainment.  However,  the  increasing  use  of  images  in  severalapplications also poses a risk of their unauthorised usage or mod-ification  without  proper  attribution  to  the  owner.  To  overcomethis issue while ensuring one-time password (OTP)–based systemauthentication,  this  study  designed  a  highly  secure  healthcareimaging  system  with  multimodal  biometrics,  data  hiding  andencryption in a deep learning environment. First, we segmenteda   medical   image   via   a   customised,   deep   neural   network   tolocate the lesion and non-lesion areas.  Next, the lesion part wasembedded into the non-lesion part via least significant bit (LSB)substitution and timestamp. Furthermore, the marked non-lesionand  lesion  parts  were  combined  to  generate  the  marked  image.Second,  encoded  multimodal  biometric  features,  i.e.  face  andiris,  and  a  novel  2D  chaotic  system  were  used  to  encrypt  themarked  image  before  transmission  over  the  network.  Throughsimulation  findings  on  security  and  accuracy  of  segmentationand  feature  extraction  design,  we  demonstrated  the  feasibilityand  effectiveness  of  our  proposed  secure  system,  highlightingtheir  superior  performance  compared  to  existing  techniques.