Improvement to the Histogram Analysis in Radiation Therapy (HART): An Open Source Software System. A. Pyakuryal* 1,2, W. K. Myint1,3, M. Gopalakrishnan1, S. Jang4, V. Sathiaseelan1,3, J. Logemann3 and B. B. Mittal1,3. (1) Northwestern Memorial Hospital, Chicago, IL. (2) University of Illinois at Chicago. Chicago, IL. (3) Northwestern University, Chicago, IL. (4) Rhode Island Hospital/ Brown University Medical Center, Providence, RI.

*Presenting Author 51st Annual Meeting of American Association of Physicists in Medicine (2009), Anaheim, CA.

ABSTRACT Purpose: The histogram analysis in radiation therapy (HART) software has been widely used for the research in intensity modulation radiation therapy (IMRT) treatments in cancer. The common application of HART is the precise and efficient dose volume histogram (DVH) analysis of structures in IMRT plans(Med Phys.35(6), p.2812 (2008)). The tool has been further developed with additional features, such as multi-dimensional dose histogram (MDH) computational module, dose response modeling (DRM), treatment plan-indices (TPIs) evaluation techniques, and biological modeling based outcome analysis (BOA) options.

Methods and Materials: Matlab based codes were designed to read RTOG data formats exported from the Pinnacle3 treatment planning system (TPS; Philips Healthcare, Best, Netherlands), and to write into a simpler HART format. HART computes the MDH differential data utilizing the information on the dose-grid values and the co-ordinates of a given structure in the TPS. MDH is applicable to evaluate spatial DVH and dose-surface histogram (DSH). The DRM utilizes the polynomial models for cumulative DVH in order to simulate the optimal dose response models for structures. The TPI assesses IMRT and stereotactic radiosurgery (SRS) plans by evaluating 10 different plan-indices as reported in literature. Similarly the POA feature can be used for evaluations of IMRT plans using various biological modeling (NTCP and TCP). DVH analysis results extracted by HART, can also be exported into customizable output formats.

Results: HART offers MDH computational capability, DRM simulations, a noble TPI evaluation technique, a simpler POA feature, and the DVH analysis module for IMRT plans. MDH computations and DRM simulations for an IMRT plan were accomplished relatively in 15-30 minutes with the clock speed of 1.8 GHz and 2 GB RAM support. The MDH and DVH analysis results were validated with the Pinnacle3 data.

Conclusions: Several applications have been incorporated into a simpler, user-friendly, and automated software package (HART). We have also implemented an open-source mechanism for various users. We expect to develop HART for various applications in radiotherapy research, and its expansion to other TPSs.

HART  OBJECTIVE : To develop an efficient computational software tool (HART) for data extraction and analysis in radiotherapy research.

HART  INTRODUCTION  Intensity modulation radiation therapy (IMRT) and 3D conformal treatment plan evaluation tools:  2D planar isodose curves .

 Orthogonal planes and the 3D isodose surfaces (Transverse, Saggital, Coronal).  Quantitative dose-volume statistics (min, max, mean, median, D95, V95) of treatment plan.  Spatial differential dose-volume histograms (sDVH), and cumulative dose-volume histogram analysis (DVH).

HART  Origination of HART :  Efficient and accurate DVH data analysis of a large number of head and neck cancer patients treated with IMRT for swallowing physiological study.  DVH based outcome analysis, histogram simulations and the plan evaluation of various IMRT treatment schemes utilizing more than 4200 DVH data-statistics per patient.  Correlation study of organ failures with the escalation of hot and cold spots in treatment plans, and its effects.  Necessity of an automated and a user friendly computational software system in radiotherapy research.

HART  METHODS : Treatment planning system: • Pinnacle3: Inverse Treatment Plans. • Sequential IMRT Boost Techniques (SqIB). • IMRT plans (Head & neck cancer): (PTV1, PTV2, PTV3, COMPOSITE). • Prescription doses: (39-46 Gy, 10-15 Gy, 10-24 Gy, Cum. dose : 64-75 Gy, Sz: 1.5 Gy / fxn). • IMRT plans ( Prostate cancer ): (PTV1, PTV2, PTV3, 4 FLD PELVIS).

• Prescription doses : (9-14 Gy, 9-10.8 Gy, 3.6-6 Gy,45 Gy, Cum. dose : 66.6- 75.8 Gy, Sz: 1.8 Gy/ fxn).

HART  HART Flow chart : TREATMENT PLANNING SYSTEM (TPS) AAPM RTOG DATA FILES

HART

PLAN INDICES EVALUATION

SPATIAL DVH ANALYSIS

MULTI-DIMENSIONAL HISTOGRAM

OUTCOME ANALYSIS

DOSE SURFACE HISTOGRAM ANALYSIS

DOSE VOLUME HISTOGRAM ANALYSIS

 DOSE VOLUME HISTOGRAM (DVH) Computation:

HART  RESULTS :  Histogram Analysis in Radiation Therapy: o To investigate the outcomes of treatment plans. o Open-source system (RTOG format compatible). o An automated & user-friendly software package. o Features :

    

Multi dimensional histogram analysis (MDH). Dose response modeling (DRM). Treatment plan-indices (TPI) evaluations. Plan specific outcome analysis (POA). DVH analysis module (DVH).

HART : Results  HART based graphical output features: PTVs Stair steps

Critical Organs End points.

Fig 1. HART simulation of cumulative DVHs for multiple structures.

HART  Customized spread sheet output features:

HART

HART  Dose response modeling (DRM) features : Differential DVH for the Gross Tumor Target (GTV)

Absolute volume (cc)

1.5 dDVH data points 1

0.5

0

0

1000

2000

3000

4000

5000

6000

Dose (cGy)

Absolute Volume (cc)

Cumulative DVH for the Gross Tumor Target (GTV) cDVH data points

Polynomial fitting

30 20 10 0

0

1000

2000

3000

4000

5000

Dose (cGy)

Fig 2. Polynomial fitting of a typical target in an IMRT plan.

HART 

Treatment plan-indices (TPI) evaluation: Quality factor (QF)

3

UPI model for plan indices Quality factor of plan: 0.99

2.5 2 1.5 1 0.5 0 0

5

10

15

25

30

UPI function

1.5

UPI score of a plan

20

COSI RCI PITV HI CN TVR DG NCI MHI TCI

1

0.5

0 0

1

2

3

4

5

6

7

8

9

10

Plan index numbers

Fig 3. Plan-indices evaluation of an IMRT plan for a prostate cancer.

HART Outcome analysis (TCP/NTCP models) features: 1

PTV1 dose volume response Poisson TCP Index = 0.22

TCP (%)

0.8

PTV2 dose volume response 0.6

Poisson TCP Index = 0.18

0.4

PTV3 dose volume response Poisson TCP Index = 0.08

0.2

GTV dose volume response Poisson TCP Index = 0.59

0 0

20

40

60

80

100

Dose (Gy) 1

PTV1 dose volume response EUD NTCP Index = 0.00

NTCP (%)

0.8

PTV2 dose volume response 0.6

EUD NTCP Index = 0.00 PTV3 dose volume response

0.4

EUD NTCP Index = 0.00 0.2

4 Field dose volume response EUD NTCP Index = 0.01

0 0

20

40

60

80

100

Dose (Gy)

Fig 4. Biological modeling for the dose response in targets (PTV, GTV) and rectum in a prostate cancer treatment.

HART  Spatial DVH (zDVH) and dose surface histogram (DSH): 1

Dose points > 100 % PD Dose points > 90 % PD Dose points > 105 % PD

Normalized volume

0.8

0.6

0.4

0.2

0 65

70

75

80

85

Slice position in z-axis

Fig 5. zDVH analysis for a typical organ in head and neck IMRT plan.

HART

 CONCLUSIONS:  HART: An in-house built open source software system.  HART: An efficient clinical tool for the IMRT based research and evaluation of treatment plans.

 HART increases the accuracy and efficiency in extracting DVH statistics from a large number of IMRT patient plans.

 HART

offers automated DVH analysis module, multidimensional histogram analysis (zDVH and DSH), dose response modeling, treatment plan-indices evaluation, and plan specific outcome analysis.

 HART is further expected to be developed as a support in radiation therapy research.

HART

 REFERENCES : 1.

2. 3. 4.

5.

S. Jang, A. Pyakuryal, K. Myint et. al., ‘A Dose Volume Histogram Analysis Software in Radiation Therapy Research’, Med Phys.35(6), p.2612 (50th AAPM Conf, Houston, Texas; 2008). J O Deasy, A I Blanco, V H Clark. CERR: A computational environment for radiotherapy research. Med Phys. 2003; 30(5): 979-985. R E Drzymala, R Mohan, L Brewster et al. DOSE VOLUME HISTOGRAMS. Int J Radiat Oncol Biol Phys.1991; 21(1): 71-78. G. Luxton, P J Keall , C R King. A new formula for normal tissue complication probability (NTCP) as a function of equivalent uniform dose (EUD). Phys Med Biol. 2008; 53: 23-36. P. Mayles, Alan Nahun et al. Handbook of Radiotherapy Physics. CRC Press,2007.

 ACKNOWLEDGEMENTS : This work was partially supported by NIH/NIDCD grant.