Dariusz Michalski, PhD

  • Clinical Assistant Professor, Radiation Oncology
  • Medical Physicist, UPMC CancerCenter

I am in charge of acquisition of 4DCT/PET scans and the evaluation of tumor motion for our lung and abdominal patients to provide tumor motion management. I am in charge of deep inspiration breath-hold CT scans for left breast cancer patients. I also supervise treatment of these patients. I also carry out monthly quality assurance of CT scanners and the RPM respiratory gating system. My research involves comparative studies of stereotactic treatment plans for CyberKnife. I am also interested in patients' respiratory dynamics during 4DCT scans.    

I am a reviewer for Medical Physics, Physics in Medicine and Biology, Radiotherapy and Technology, International Journal of Radiation Oncology Biology Physics.

Teaching Activities:

All teaching concerned background theory and practicum of the clinical application 4DCT and 4DPET technology or left breast breath hold scanning and treatment.

  • 2015-08-24 12:00-4:00 pm     UPMC Radiation Oncology Center in Erie, PA
  • 2015-09-01 1:00pm:4:00pm   UPMC Shadyside CT sim and my office Brian King Erie Physicist
  • 2015-09-29 12:00-4:00pm      UPMC Shadyside CT sim and my office Bin Hu  Erie Physicist
  • 2015-10-13 12:00-5:00pm      UPMC Shadyside CT sim and my office Elongovan Doraisamy UPMC Murtha physicists
  • 2015-11-17 2:00 pm - 7:00 pm      UPMC Shadyside CT sim and my office Frank Ottino Washington physicist
  • 2016-2-2 4:00-6:00pm      CT sim; Chris Houser UPMC Magee physicist
  • 2016-2-9, 1:00-4:00    CT sim + Dosimetry Chris Houser
  • 2016-2-29 - 2016-3-4             My office , ct sim , dosimetry 15 hours Luis Lagrase UPMC Columbia
  • 2016-06-21                             My office and ct sim. Our dosimetrist Doris Chen 3:00-4:00pm

Representative Publications

Darek Michalski, M. Saiful Huq, Greg Bednarz, and Dwight E. Heron, “The use of strain tensor to estimate thoracic tumors deformation”, Medical Physics 41, 073503 (2014); doi: 10.1118/1.4884222

Hayeon Kim, M. Saiful Huq, Chris Houser, Sushil Beriwal, Dariusz Michalski, “Mapping of dose distribution from IMRT onto MRI-guided high dose rate brachytherapy using deformable image registration for cervical cancer treatments: preliminary study with commercially available software” J Contemp Brachytherapy 2014; 6, 2: 178–184, DOI: 10.5114/jcb.2014.43240

Research Interests

Our clinic operates CyberKnife M6 that can use fixed-cones, variable aperture Iris collimators and InCisemultileaf collimator (MLC). The IRIS collimator system consists of 12 dodecagon-shaped collimators raging from 5 to 60 mm. The InCise MLC System consists of 41 leaves of 2.5 mm-wide resolution projected at 800 mm Source-to-Axis Distance. This study concerns the feasibility, dosimetric quality, and delivery efficacy of SBRT pancreatic cancer treatment plans using cone/Iris collimators and MLC on CyberKnife. This retrospective study includes 35 patients who originally underwent pancreatic Stereotactic Radiation Therapy on Varian's liniacs. The coverage is better for Iris than for MLC plans. The former had the mean coverage of 91.26%, and MLC plans have the mean coverage of 86.13%. Conformity index does not differ statistically for both plans with the mean CI for Iris 1.12 and 1.14 for MLC.  There is a significant difference in monitor units between these plans, which also results in the difference in time delivery. For Iris plans the mean monitor units is 53947.17, and the mean delivery time is 97.46 min. For MLC plans the mean monitor units is 20951.97, and the mean delivery time is 49.94 min. Optimization of the MLC plans was more challenging than Iris plans, especially with providing sparing of the bowel. Dosimetrically Iris plans are superior to MLC plans. The feasibility of the creation of SBRT pancreatic cancer plans for Iris/Cone and MLC CyberKnife was demonstrated. However from clinical perspective their delivery time can be viewed as prohibitive, particularly Iris plans and only an extraordinary clinical scenario would warrant their administration.

The intuitive view of ventilation is that of a simple exhale-inhale sinusoidal-like process. Yet its nature is complex and exhibits non-linear features. For the former, respiratory dynamics (RD) is approximated with linear parameters, which are easily identified and understood. For the latter the state space is reconstructed. Data consists of 20 respiratory tracks (RT) of lung cancer patients. Linear parameters are extracted from exhale-inhale decomposition. The non-linear analysis is based on the method of delays that unfolds the state space of RD. The stationarity of RD is examined from non-linear perspective of the RD evolution operator. The determinism is established with two independent methods. Non-linear statistics test RTs and their surrogate data. The largest Lyapunov exponents, LLE, are determined. Linear models only approximate breathing patterns. Tests on surrogate data demonstrate non-linearity. The LLEs shows RD's sensitive dependence on the initial conditions that demonstrates chaotic behavior. The findings suggest the independence of exhalation and inhalation. The state space invariants compared with intuitive characteristics of RT provide new insight into respiratory patterns and lung tumor motion management in radiation oncology. It affects patients' CT imaging, the treatment decision process and the treatment administration. Since respiratory signal, which in a sense reflects tumor motion, exhibits these features it suggests tumor motion exhibits the same characteristics. The non-linear approach leads to more realistic evaluation of respiratory motion complexity and may lead to the improvement of tumor motion management.